From amelie.serpollet at cea.fr Wed Aug 3 10:30:38 2011 From: amelie.serpollet at cea.fr (=?iso-8859-1?Q?SERPOLLET_Am=E9lie_228173?=) Date: Wed, 3 Aug 2011 08:30:38 +0000 Subject: [FieldTrip] delay with realtime buffer Message-ID: <04B8DFAA8CFBED46A48C995E5BB1CD459A39@EXDAG0-B0.intra.cea.fr> Dear Fieldtrip users and developers, Using Fieldtrip buffer for a BCI loop (in realtime), I want to measure the delay introduced by the use of the buffer. I use two simple programs to measure it : the first one sends data to a buffer and writes CPU time in a file when a rising edge was detected in the signal sent ; the other program requests the data and also writes CPU time in a file when the same rising edge is detected in the signal received. I checked that the delay introduced by the detection and file writing is insignificant (0.5 to 5 µs). On my computer, with 32 channels signal and chunks of 8 to 32 points, I measured very variables delays, also after deactivating the anti-virus : 1ms to more than 45ms. Do you know what it is due to, if I can reduce it or if I can make it less variable ? Thank you in advance. Amélie -------------- next part -------------- An HTML attachment was scrubbed... URL: From Don.Rojas at ucdenver.edu Wed Aug 3 17:47:56 2011 From: Don.Rojas at ucdenver.edu (Rojas, Don) Date: Wed, 3 Aug 2011 09:47:56 -0600 Subject: [FieldTrip] ft_volumesegment question Message-ID: <621FD28E-EB95-4966-9612-B64308F79074@ucdenver.edu> Dear Fieldtrippers, I use FieldTrip and SPM8 together quite frequently. I noticed recently that the segmentation output (gray, white and csf tpm) of ft_volumesegment (c1, c2, c3) does not align properly to the input mri scan. This is not the case when doing the same segmentation in SPM8. I'm using the 20110725 version of Fieldtrip and have mad certain that it and not the one in spm8 is used. My cfg and command is as follows: cfg.output = {'brain' 'skull' 'scalp','tpm'}; cfg.spmversion = 'spm8'; cfg.name = 'test'; cfg.write = 'yes'; cfg.coordsys = 'spm'; seg = ft_volumesegment(cfg,mri); Looking at the code for ft_volumesegment, it appears that a permutation of the mri dimensions (swapping 2 and 3) is done on line 279: [mri,permutevec,flipflags] = align_ijk2xyz(mri); The mri.transform field is changed accordingly. However, the original.transform is put into the headers for the c1, c2 and c3 output (line 399). I think that this is not correct because the segmentation output has the permuted dimensions, not the dimensions of the original mri. So, the correct transform should be the mri.transform after align_ijk2xyz I think. Or maybe this is intentional behavior and I'm missing something. It can easily be checked, however, using Check Reg, in SPM8, that the output tpm do not align correctly with the original mri. If I manually write the mri.transform (from align_ijk2xyz) to the tpm output, however, they do align correctly. Best, Don ----------------------- Don Rojas, Ph.D. Associate Professor of Psychiatry U. of Colorado Denver Anschutz Medical Campus Director, UCD Magnetoencephalography Lab 13001 E. 17th Pl F546 Aurora, CO 80045 303-724-4994 -------------- next part -------------- An HTML attachment was scrubbed... URL: From ekanal at cmu.edu Wed Aug 3 23:29:52 2011 From: ekanal at cmu.edu (Kanal Eliezer) Date: Wed, 3 Aug 2011 17:29:52 -0400 Subject: [FieldTrip] timelock grand average can't average gradiometers? Message-ID: <2788E298-FE53-4F74-9C27-FDC29C0DDCBF@cmu.edu> Hello all - I'm trying to average a number of datasets collected from a Neuromag scanner. This dataset contains both magnetometer and gradiometer data. I'm trying to use ft_timelockgrandaverage to average the cleaned data, and I'm getting the following error message: > Warning: discarding gradiometer information because it cannot be averaged My question is, why not? My gradiometer data isn't different than the magnetometer data. Even more confusing, when looking in the code, it looks like it's averaging it anyway, and the resulting dataset seems to contain averaged grad data. What's the purpose of this error? Thanks, Eliezer Kanal From eelke.spaak at donders.ru.nl Thu Aug 4 08:03:32 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Thu, 4 Aug 2011 08:03:32 +0200 Subject: [FieldTrip] timelock grand average can't average gradiometers? In-Reply-To: <2788E298-FE53-4F74-9C27-FDC29C0DDCBF@cmu.edu> References: <2788E298-FE53-4F74-9C27-FDC29C0DDCBF@cmu.edu> Message-ID: Hi Eli, The warning message you mention regards the information about the gradiometer positions, i.e. the information in the data.grad field. The MEG-data recorded by the gradiometer sensors will still be averaged. I understand the confusion, though, and agree that the message could be a bit clearer. Best, Eelke 2011/8/3 Kanal Eliezer : > Hello all - > > I'm trying to average a number of datasets collected from a Neuromag scanner. This dataset contains both magnetometer and gradiometer data. I'm trying to use ft_timelockgrandaverage to average the cleaned data, and I'm getting the following error message: > >> Warning: discarding gradiometer information because it cannot be averaged > > My question is, why not? My gradiometer data isn't different than the magnetometer data. Even more confusing, when looking in the code, it looks like it's averaging it anyway, and the resulting dataset seems to contain averaged grad data. What's the purpose of this error? > > Thanks, > Eliezer Kanal > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From anna.lambrechts at gmail.com Thu Aug 4 11:17:46 2011 From: anna.lambrechts at gmail.com (Anna Lambrechts) Date: Thu, 4 Aug 2011 11:17:46 +0200 Subject: [FieldTrip] fieldtrip Digest, Vol 8, Issue 40 In-Reply-To: References: Message-ID: Hi, thank you for your first answer. Actually my question is really on the right way to implement a group comparison in fieldtrip (patient vs control group). Is it ok to run an independent t-test with the following design: design = zeros(2,n_group1+n_group2); design(2,:) = 1; design(1,1:n_group1) = 1:n_group1; design(1,n_group1+1:end) = 1:n_group2; A subsequent question is whether it is possible to implement a mixed design or we always have to proceed by combining conditions and compare it between groups. Cheers, Anna. > Message: 2 > Date: Sat, 30 Jul 2011 20:45:12 +0200 > From: "Eric Maris" > To: "'Email discussion list for the FieldTrip project'" > > Subject: Re: [FieldTrip] Group comparison statistics - ERF study > Message-ID: <03a601cc4ee8$d152b760$73f82620$@maris at donders.ru.nl> > Content-Type: text/plain; charset="us-ascii" > > Hi Anna, > > > > You can compare the two groups with respect to any linear combination that > can be formed using the observations in the 2 (conditions) x 2 (response > types) within-UO design. For testing interaction effects, these linear > combinations are the usual contrasts. > > > > > > Best, > > > > Eric Maris > > > > > > dr. Eric Maris > Donders Institute for Brain, Cognition and Behavior > > Radboud University > P.O. Box 9104 > 6500 HE Nijmegen > The Netherlands > T:+31 24 3612651 > Mobile: 06 39584581 > > F:+31 24 3616066 > mailto:e.maris at donders.ru.nl > > http://www.nphyscog.com/ > > > > > > > > From: fieldtrip-bounces at donders.ru.nl > [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Anna Lambrechts > Sent: donderdag 28 juli 2011 11:32 > To: fieldtrip at donders.ru.nl > Subject: [FieldTrip] Group comparison statistics - ERF study > > > > Hi, > > I am trying to run a group comparison analysis on event-related fields data > in a 2 (groups) x 2 (conditions) x 2 (response types) design. Is this > possible at all with any fieldtrip script? As far as I know implemented > statistics look at within-groups comparison. > > Thanks, > Anna. > > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: < > http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20110730/1f410391/attachment-0001.html > > > > ------------------------------ > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > End of fieldtrip Digest, Vol 8, Issue 40 > **************************************** > -------------- next part -------------- An HTML attachment was scrubbed... URL: From v.litvak at ion.ucl.ac.uk Thu Aug 4 12:39:53 2011 From: v.litvak at ion.ucl.ac.uk (Vladimir Litvak) Date: Thu, 4 Aug 2011 11:39:53 +0100 Subject: [FieldTrip] timelock grand average can't average gradiometers? In-Reply-To: References: <2788E298-FE53-4F74-9C27-FDC29C0DDCBF@cmu.edu> Message-ID: Dear Eelke, Actually there is a way to average sensors which is implemented in ft_average_sens. Perhaps it would be a good idea to use it in ft_timelockgrandaverage. Best, Vladimir On Thu, Aug 4, 2011 at 7:03 AM, Eelke Spaak wrote: > Hi Eli, > > The warning message you mention regards the information about the > gradiometer positions, i.e. the information in the data.grad field. > The MEG-data recorded by the gradiometer sensors will still be > averaged. > > I understand the confusion, though, and agree that the message could > be a bit clearer. > > Best, > Eelke > > 2011/8/3 Kanal Eliezer : >> Hello all - >> >> I'm trying to average a number of datasets collected from a Neuromag scanner. This dataset contains both magnetometer and gradiometer data. I'm trying to use ft_timelockgrandaverage to average the cleaned data, and I'm getting the following error message: >> >>> Warning: discarding gradiometer information because it cannot be averaged >> >> My question is, why not? My gradiometer data isn't different than the magnetometer data. Even more confusing, when looking in the code, it looks like it's averaging it anyway, and the resulting dataset seems to contain averaged grad data. What's the purpose of this error? >> >> Thanks, >> Eliezer Kanal >> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From eelke.spaak at donders.ru.nl Thu Aug 4 14:04:44 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Thu, 4 Aug 2011 14:04:44 +0200 Subject: [FieldTrip] timelock grand average can't average gradiometers? In-Reply-To: References: <2788E298-FE53-4F74-9C27-FDC29C0DDCBF@cmu.edu> Message-ID: Dear Vladimir, Thanks, I did not know about that function. However, I don't think ft_timelockgrandaverage should automatically average sensor positions; if the user wants this it is probably best if he/she does this explicitly. But, I will bring it up in the next FieldTrip dev team meeting, and see what others think (particularly Robert and Jan-Mathijs). Best, Eelke 2011/8/4 Vladimir Litvak : > Dear Eelke, > > Actually there is a way to average sensors which is implemented in > ft_average_sens. Perhaps it would be a good idea to use it in > ft_timelockgrandaverage. > > Best, > > Vladimir > > On Thu, Aug 4, 2011 at 7:03 AM, Eelke Spaak wrote: >> Hi Eli, >> >> The warning message you mention regards the information about the >> gradiometer positions, i.e. the information in the data.grad field. >> The MEG-data recorded by the gradiometer sensors will still be >> averaged. >> >> I understand the confusion, though, and agree that the message could >> be a bit clearer. >> >> Best, >> Eelke >> >> 2011/8/3 Kanal Eliezer : >>> Hello all - >>> >>> I'm trying to average a number of datasets collected from a Neuromag scanner. This dataset contains both magnetometer and gradiometer data. I'm trying to use ft_timelockgrandaverage to average the cleaned data, and I'm getting the following error message: >>> >>>> Warning: discarding gradiometer information because it cannot be averaged >>> >>> My question is, why not? My gradiometer data isn't different than the magnetometer data. Even more confusing, when looking in the code, it looks like it's averaging it anyway, and the resulting dataset seems to contain averaged grad data. What's the purpose of this error? >>> >>> Thanks, >>> Eliezer Kanal >>> >>> >>> _______________________________________________ >>> fieldtrip mailing list >>> fieldtrip at donders.ru.nl >>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >>> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From e.maris at donders.ru.nl Thu Aug 4 14:08:39 2011 From: e.maris at donders.ru.nl (Eric Maris) Date: Thu, 4 Aug 2011 14:08:39 +0200 Subject: [FieldTrip] fieldtrip Digest, Vol 8, Issue 40 In-Reply-To: References: Message-ID: <053f01cc529f$3efff020$bcffd060$@maris@donders.ru.nl> Hi Anna, thank you for your first answer. Actually my question is really on the right way to implement a group comparison in fieldtrip (patient vs control group). Is it ok to run an independent t-test with the following design: design = zeros(2,n_group1+n_group2); design(2,:) = 1; design(1,1:n_group1) = 1:n_group1; design(1,n_group1+1:end) = 1:n_group2; This is not correct. I advise you to have a look at the Fieldtrip tutorial on cluster-based permutation tests. A between-subjects experiment (e.g., patients versus controls) is formally equivalent to a between-trials experiment (explained in detail in the tutorial), and therefore should be analyzed in the same way. You can even copy-paste part of the tutorial code for the analysis of your between-subjects data. A subsequent question is whether it is possible to implement a mixed design or we always have to proceed by combining conditions and compare it between groups. I do not see the contrast between these two options. What is wrong with the second option? In fact, it is the option that is advocated in many statistics books (e.g., Maxwell & Delaney, Johnson & Wichren). Best, Eric Maris Cheers, Anna. Message: 2 Date: Sat, 30 Jul 2011 20:45:12 +0200 From: "Eric Maris" To: "'Email discussion list for the FieldTrip project'" Subject: Re: [FieldTrip] Group comparison statistics - ERF study Message-ID: <03a601cc4ee8$d152b760$73f82620$@maris at donders.ru.nl> Content-Type: text/plain; charset="us-ascii" Hi Anna, You can compare the two groups with respect to any linear combination that can be formed using the observations in the 2 (conditions) x 2 (response types) within-UO design. For testing interaction effects, these linear combinations are the usual contrasts. Best, Eric Maris dr. Eric Maris Donders Institute for Brain, Cognition and Behavior Radboud University P.O. Box 9104 6500 HE Nijmegen The Netherlands T:+31 24 3612651 Mobile: 06 39584581 F:+31 24 3616066 mailto:e.maris at donders.ru.nl http://www.nphyscog.com/ From: fieldtrip-bounces at donders.ru.nl [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Anna Lambrechts Sent: donderdag 28 juli 2011 11:32 To: fieldtrip at donders.ru.nl Subject: [FieldTrip] Group comparison statistics - ERF study Hi, I am trying to run a group comparison analysis on event-related fields data in a 2 (groups) x 2 (conditions) x 2 (response types) design. Is this possible at all with any fieldtrip script? As far as I know implemented statistics look at within-groups comparison. Thanks, Anna. -------------- next part -------------- An HTML attachment was scrubbed... URL: ------------------------------ _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip End of fieldtrip Digest, Vol 8, Issue 40 **************************************** -------------- next part -------------- An HTML attachment was scrubbed... URL: From drivolta81 at gmail.com Thu Aug 4 14:44:08 2011 From: drivolta81 at gmail.com (Davide Rivolta) Date: Thu, 4 Aug 2011 14:44:08 +0200 Subject: [FieldTrip] Potential bug in trl Message-ID: Dear all, I have noticed that the trial definition works well when begsample, ensample and offset have fixed values. I have however a specific trialfunction, where each trial has a different lenght. This works well when you specify a fixed offset, however, when even the offset has to change for each trial, Fieldtrip reports values until 0. It ignores time after 0. Might it be a bug? Thanks, Davide -------------- next part -------------- An HTML attachment was scrubbed... URL: From BelluscB at ninds.nih.gov Thu Aug 4 15:23:57 2011 From: BelluscB at ninds.nih.gov (Belluscio, Beth (NIH/NINDS) [E]) Date: Thu, 4 Aug 2011 09:23:57 -0400 Subject: [FieldTrip] time frequency analysis - basic question Message-ID: <794DFDD3C128EF44AC49ADC58FD0090C057456B617@NIHMLBX10.nih.gov> Hi all- When calling ft_freqanalysis you specify the foi limits and intervals - eg 5:5:30. The resulting data set appears to generate the power at discreet frequencies (5, 10, 15, 20, 25, 30). Does this actually reflect an estimate of the power over a range of frequencies (2.5-7.5, 7.5-12.5, 12.5-17.5,....) or the power at discreet frequencies with those in between not evaluated? Thanks for your help, Beth. Beth Belluscio MD-PhD Clinical Fellow Human Motor Control Section NINDS, NIH 301-402-3495 -------------- next part -------------- An HTML attachment was scrubbed... URL: From michael.wibral at web.de Thu Aug 4 15:55:46 2011 From: michael.wibral at web.de (Michael Wibral) Date: Thu, 4 Aug 2011 15:55:46 +0200 (CEST) Subject: [FieldTrip] Potential bug in trl Message-ID: <422581888.3319195.1312466146646.JavaMail.fmail@mwmweb086> An HTML attachment was scrubbed... URL: From drivolta81 at gmail.com Thu Aug 4 16:03:24 2011 From: drivolta81 at gmail.com (Davide Rivolta) Date: Thu, 4 Aug 2011 16:03:24 +0200 Subject: [FieldTrip] Potential bug in trl In-Reply-To: <422581888.3319195.1312466146646.JavaMail.fmail@mwmweb086> References: <422581888.3319195.1312466146646.JavaMail.fmail@mwmweb086> Message-ID: Thanks Michael, I wrote my own trialfun. The offset would be negative and would have the same lenght of difference between endsample-begsample. This, for each trial, would have a different value. It can be a bug. In fact, I tried to vhange the trialfun by leaving the same begsample and endsample, but I put a fixed offset (i.e., -100) and it worked. All the trial was in and not, like with the variable offset, until time = 0. Thanks, Davide On Thu, Aug 4, 2011 at 3:55 PM, Michael Wibral wrote: > Hi davide, > > I guess in such a case you would have to write your own trial function. > However, offset values larger than 0 mean, that the stimulus occured before > you start your piece of data - is that really what you wanted ? > > Michael > > > ------------------------------ > *Von:* "Davide Rivolta" > *Gesendet:* Aug 4, 2011 2:44:08 PM > *An:* fieldtrip at donders.ru.nl > *Betreff:* [FieldTrip] Potential bug in trl > > > Dear all, > > I have noticed that the trial definition works well when begsample, > ensample and offset have fixed values. > > I have however a specific trialfunction, where each trial has a different > lenght. > > This works well when you specify a fixed offset, however, when even the > offset has to change for each trial, Fieldtrip reports values until 0. It > ignores time after 0. > > > Might it be a bug? > > Thanks, > > Davide > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > -- Davide Rivolta, PhD -------------- next part -------------- An HTML attachment was scrubbed... URL: From jm.horschig at donders.ru.nl Fri Aug 5 09:13:23 2011 From: jm.horschig at donders.ru.nl (=?ISO-8859-1?Q?=22J=F6rn_M=2E_Horschig=22?=) Date: Fri, 05 Aug 2011 09:13:23 +0200 Subject: [FieldTrip] Potential bug in trl In-Reply-To: References: <422581888.3319195.1312466146646.JavaMail.fmail@mwmweb086> Message-ID: <4E3B9813.8070507@donders.ru.nl> Dear Davide, I never worked with trials of different length, however this should be straight forward when writing your own trialfun (just as Michael proposed). In most trialfuns, the trl matrix is constructed by concatenating the vectors begsample, endsample and offset. These vectors can have any values you want them to have. As a simple example, you can write a trialfun that just returns trl = [1 2 1; 400 500 100] (for your experiment, you of course want these values to be computed from your trigger values). Is your problem that such a thing does not work? It would be great if you give a concrete use-case, or e.g. attach your trialfun. Best, Jörn On 8/4/2011 4:03 PM, Davide Rivolta wrote: > Thanks Michael, > I wrote my own trialfun. > The offset would be negative and would have the same lenght of > difference between endsample-begsample. This, for each trial, would > have a different value. > It can be a bug. In fact, I tried to vhange the trialfun by leaving > the same begsample and endsample, but I put a fixed offset (i.e., > -100) and it worked. All the trial was in and not, like with the > variable offset, until time = 0. > Thanks, > Davide > > On Thu, Aug 4, 2011 at 3:55 PM, Michael Wibral > wrote: > > Hi davide, > > I guess in such a case you would have to write your own trial > function. However, offset values larger than 0 mean, that the > stimulus occured before you start your piece of data - is that > really what you wanted ? > > Michael > > > ------------------------------------------------------------------------ > *Von:* "Davide Rivolta" > > *Gesendet:* Aug 4, 2011 2:44:08 PM > *An:* fieldtrip at donders.ru.nl > *Betreff:* [FieldTrip] Potential bug in trl > > > Dear all, > I have noticed that the trial definition works well when > begsample, ensample and offset have fixed values. > I have however a specific trialfunction, where each trial has > a different lenght. > This works well when you specify a fixed offset, however, when > even the offset has to change for each trial, Fieldtrip > reports values until 0. It ignores time after 0. > Might it be a bug? > Thanks, > Davide > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > > -- > Davide Rivolta, PhD > > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip -- Jörn M. Horschig PhD Student Donders Institute for Brain, Cognition and Behaviour Centre for Cognitive Neuroimaging Radboud University Nijmegen Neuronal Oscillations Group P.O. Box 9101 NL-6500 HB Nijmegen The Netherlands Contact: E-Mail: jm.horschig at donders.ru.nl Tel: +31-(0)24-36-68493 Web: http://www.ru.nl/donders Visiting address: Trigon, room 2.30 Kapittelweg 29 NL-6525 EN Nijmegen The Netherlands -------------- next part -------------- An HTML attachment was scrubbed... URL: From jm.horschig at donders.ru.nl Fri Aug 5 09:29:38 2011 From: jm.horschig at donders.ru.nl (=?ISO-8859-1?Q?=22J=F6rn_M=2E_Horschig=22?=) Date: Fri, 05 Aug 2011 09:29:38 +0200 Subject: [FieldTrip] time frequency analysis - basic question In-Reply-To: <794DFDD3C128EF44AC49ADC58FD0090C057456B617@NIHMLBX10.nih.gov> References: <794DFDD3C128EF44AC49ADC58FD0090C057456B617@NIHMLBX10.nih.gov> Message-ID: <4E3B9BE2.4000300@donders.ru.nl> Dear Beth, From all my knowledge I gathered so far, I can try to answer your question (and hope that people from this list correct me if I am wrong): As often, the truth lies somewhere in between. The frequency smoothing is determined by the type and length of the taper you apply - this is similar to wavelet analysis. If you use multitapers, you can specify cfg.tapsmofrq, which determines the frequency band you want to smooth with. The FT freqanalysis method uses multiplication in the time domain instead of convolution in the frequency domain (which is equivalent). Any taper will always have a specific frequency spectrum itself, and thus it will not give you the value of a single frequency (though the output looks as if). So if you specify foi as [5 10 15], then the power at 5Hz will not be from 2.5 to 7.5Hz, but neither will it be at exactly and only 5Hz (btw, the exact center frequency might not be 5Hz, this depends on the (Rayleigh) frequency resolution of your signal). I do not know any more specifics concerning this, maybe the help of ft_freqanalysis can tell you some additonal things: http://fieldtrip.fcdonders.nl/reference/ft_freqanalysis Best, Jörn Btw, if this is not correct, else from the mailinglist should correct me asap ;) On 8/4/2011 3:23 PM, Belluscio, Beth (NIH/NINDS) [E] wrote: > > Hi all- > > When calling ft_freqanalysis you specify the foi limits and > intervals -- eg 5:5:30. The resulting data set appears to generate > the power at discreet frequencies (5, 10, 15, 20, 25, 30). Does this > actually reflect an estimate of the power over a range of frequencies > (2.5-7.5, 7.5-12.5, 12.5-17.5,....) or the power at discreet > frequencies with those in between not evaluated? > > Thanks for your help, > > Beth. > > Beth Belluscio MD-PhD > > Clinical Fellow > > Human Motor Control Section > > NINDS, NIH > > 301-402-3495 > > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip -- Jörn M. Horschig PhD Student Donders Institute for Brain, Cognition and Behaviour Centre for Cognitive Neuroimaging Radboud University Nijmegen Neuronal Oscillations Group P.O. Box 9101 NL-6500 HB Nijmegen The Netherlands Contact: E-Mail: jm.horschig at donders.ru.nl Tel: +31-(0)24-36-68493 Web: http://www.ru.nl/donders Visiting address: Trigon, room 2.30 Kapittelweg 29 NL-6525 EN Nijmegen The Netherlands -------------- next part -------------- An HTML attachment was scrubbed... URL: From drivolta81 at gmail.com Fri Aug 5 11:36:07 2011 From: drivolta81 at gmail.com (Davide Rivolta) Date: Fri, 5 Aug 2011 11:36:07 +0200 Subject: [FieldTrip] Potential bug in trl In-Reply-To: <4E3B9813.8070507@donders.ru.nl> References: <422581888.3319195.1312466146646.JavaMail.fmail@mwmweb086> <4E3B9813.8070507@donders.ru.nl> Message-ID: Dear Jörn, Thank you for your reply. I guess I have now figured it out. Thanks, Davide On Fri, Aug 5, 2011 at 9:13 AM, "Jörn M. Horschig" < jm.horschig at donders.ru.nl> wrote: > Dear Davide, > > I never worked with trials of different length, however this should be > straight forward when writing your own trialfun (just as Michael proposed). > In most trialfuns, the trl matrix is constructed by concatenating the > vectors begsample, endsample and offset. These vectors can have any values > you want them to have. As a simple example, you can write a trialfun that > just returns trl = [1 2 1; 400 500 100] (for your experiment, you of course > want these values to be computed from your trigger values). > Is your problem that such a thing does not work? It would be great if you > give a concrete use-case, or e.g. attach your trialfun. > > Best, > Jörn > > > On 8/4/2011 4:03 PM, Davide Rivolta wrote: > > Thanks Michael, > > I wrote my own trialfun. > > The offset would be negative and would have the same lenght of difference > between endsample-begsample. This, for each trial, would have a different > value. > > It can be a bug. In fact, I tried to vhange the trialfun by leaving the > same begsample and endsample, but I put a fixed offset (i.e., -100) and it > worked. All the trial was in and not, like with the variable offset, until > time = 0. > > Thanks, > > Davide > > On Thu, Aug 4, 2011 at 3:55 PM, Michael Wibral wrote: > >> Hi davide, >> >> I guess in such a case you would have to write your own trial function. >> However, offset values larger than 0 mean, that the stimulus occured before >> you start your piece of data - is that really what you wanted ? >> >> Michael >> >> >> ------------------------------ >> *Von:* "Davide Rivolta" >> *Gesendet:* Aug 4, 2011 2:44:08 PM >> *An:* fieldtrip at donders.ru.nl >> *Betreff:* [FieldTrip] Potential bug in trl >> >> >> Dear all, >> >> I have noticed that the trial definition works well when begsample, >> ensample and offset have fixed values. >> >> I have however a specific trialfunction, where each trial has a different >> lenght. >> >> This works well when you specify a fixed offset, however, when even the >> offset has to change for each trial, Fieldtrip reports values until 0. It >> ignores time after 0. >> >> >> Might it be a bug? >> >> Thanks, >> >> Davide >> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> > > > > -- > Davide Rivolta, PhD > > > > _______________________________________________ > fieldtrip mailing listfieldtrip at donders.ru.nlhttp://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > -- > Jörn M. Horschig > PhD Student > Donders Institute for Brain, Cognition and Behaviour > Centre for Cognitive Neuroimaging > Radboud University Nijmegen > Neuronal Oscillations Group > > P.O. Box 9101 > NL-6500 HB Nijmegen > The Netherlands > > Contact: > E-Mail: jm.horschig at donders.ru.nl > Tel: +31-(0)24-36-68493 > Web: http://www.ru.nl/donders > > Visiting address: > Trigon, room 2.30 > Kapittelweg 29 > NL-6525 EN Nijmegen > The Netherlands > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > -- Davide Rivolta, PhD -------------- next part -------------- An HTML attachment was scrubbed... URL: From BelluscB at ninds.nih.gov Fri Aug 5 16:20:11 2011 From: BelluscB at ninds.nih.gov (Belluscio, Beth (NIH/NINDS) [E]) Date: Fri, 5 Aug 2011 10:20:11 -0400 Subject: [FieldTrip] time frequency analysis - basic question In-Reply-To: <4E3B9BE2.4000300@donders.ru.nl> References: <794DFDD3C128EF44AC49ADC58FD0090C057456B617@NIHMLBX10.nih.gov> <4E3B9BE2.4000300@donders.ru.nl> Message-ID: <794DFDD3C128EF44AC49ADC58FD0090C057456B626@NIHMLBX10.nih.gov> Thanks, Jorn, that makes sense to me. From: "Jörn M. Horschig" [mailto:jm.horschig at donders.ru.nl] Sent: Friday, August 05, 2011 3:30 AM To: Email discussion list for the FieldTrip project Subject: Re: [FieldTrip] time frequency analysis - basic question Dear Beth, >From all my knowledge I gathered so far, I can try to answer your question (and hope that people from this list correct me if I am wrong): As often, the truth lies somewhere in between. The frequency smoothing is determined by the type and length of the taper you apply - this is similar to wavelet analysis. If you use multitapers, you can specify cfg.tapsmofrq, which determines the frequency band you want to smooth with. The FT freqanalysis method uses multiplication in the time domain instead of convolution in the frequency domain (which is equivalent). Any taper will always have a specific frequency spectrum itself, and thus it will not give you the value of a single frequency (though the output looks as if). So if you specify foi as [5 10 15], then the power at 5Hz will not be from 2.5 to 7.5Hz, but neither will it be at exactly and only 5Hz (btw, the exact center frequency might not be 5Hz, this depends on the (Rayleigh) frequency resolution of your signal). I do not know any more specifics concerning this, maybe the help of ft_freqanalysis can tell you some additonal things: http://fieldtrip.fcdonders.nl/reference/ft_freqanalysis Best, Jörn Btw, if this is not correct, else from the mailinglist should correct me asap ;) On 8/4/2011 3:23 PM, Belluscio, Beth (NIH/NINDS) [E] wrote: Hi all- When calling ft_freqanalysis you specify the foi limits and intervals - eg 5:5:30. The resulting data set appears to generate the power at discreet frequencies (5, 10, 15, 20, 25, 30). Does this actually reflect an estimate of the power over a range of frequencies (2.5-7.5, 7.5-12.5, 12.5-17.5,....) or the power at discreet frequencies with those in between not evaluated? Thanks for your help, Beth. Beth Belluscio MD-PhD Clinical Fellow Human Motor Control Section NINDS, NIH 301-402-3495 _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip -- Jörn M. Horschig PhD Student Donders Institute for Brain, Cognition and Behaviour Centre for Cognitive Neuroimaging Radboud University Nijmegen Neuronal Oscillations Group P.O. Box 9101 NL-6500 HB Nijmegen The Netherlands Contact: E-Mail: jm.horschig at donders.ru.nl Tel: +31-(0)24-36-68493 Web: http://www.ru.nl/donders Visiting address: Trigon, room 2.30 Kapittelweg 29 NL-6525 EN Nijmegen The Netherlands -------------- next part -------------- An HTML attachment was scrubbed... URL: From cas243 at georgetown.edu Sat Aug 6 17:21:47 2011 From: cas243 at georgetown.edu (Clara A. Scholl) Date: Sat, 6 Aug 2011 11:21:47 -0400 Subject: [FieldTrip] Extracting a timecourse from cluster of channels In-Reply-To: <4e34099a.c188e50a.3fc3.ffff8898SMTPIN_ADDED@mx.google.com> References: <4e34099a.c188e50a.3fc3.ffff8898SMTPIN_ADDED@mx.google.com> Message-ID: Thanks Eric! Do you think it's reasonable to take the intersection, or union, of channels over time in a given space-time cluster? Do you know of any publications that have dealt with this issue (choosing channels from a space-time cluster for timecourse extraction) that I could look at & reference? Thanks, Clara On Sat, Jul 30, 2011 at 9:31 AM, Eric Maris wrote: > Dear Clara, > >> I am using fieldtrip's cluster-based permutation test to identify a >> significant channel-time cluster showing an effect.  Now I would like >> to extract a timecourse averaged over these channels (for the purpose >> of comparing different conditions, not used to generate the cluster), >> but I'm not sure if it makes sense to do so because the channels >> belonging to the cluster change over time -- would the timecourse have >> contributions from different channels at different time points?  Do I >> need to limit my cluster to fixed channels across time? (And if so, >> how do I do that?) > > Yes, you should fix channels across time. The most important point in > selecting the channels is finding evidence for the fact that the channels > reflects a single source only. This is not a statistical issue, and > therefore there is not a p-value that can guide you in this choice. If you > have collected MEG data, the best evidence is a dipolar topography of your > effect. > > Best, > > Eric Maris > > >> >> Thanks! >> Clara >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From jan.schoffelen at donders.ru.nl Sat Aug 6 21:08:02 2011 From: jan.schoffelen at donders.ru.nl (jan-mathijs schoffelen) Date: Sat, 6 Aug 2011 21:08:02 +0200 Subject: [FieldTrip] ft_volumesegment question In-Reply-To: <621FD28E-EB95-4966-9612-B64308F79074@ucdenver.edu> References: <621FD28E-EB95-4966-9612-B64308F79074@ucdenver.edu> Message-ID: Dear Don, You're absolutely right. The correct tranfsormation matrix written to the file should be the mri.transform, not the original.transform. I'll fix this as soon as possible. Thanks for the debugging. Best wishes, Jan-Mathijs On Aug 3, 2011, at 5:47 PM, Rojas, Don wrote: > Dear Fieldtrippers, > > I use FieldTrip and SPM8 together quite frequently. I noticed recently that the segmentation output (gray, white and csf tpm) of ft_volumesegment (c1, c2, c3) does not align properly to the input mri scan. This is not the case when doing the same segmentation in SPM8. I'm using the 20110725 version of Fieldtrip and have mad certain that it and not the one in spm8 is used. My cfg and command is as follows: > > cfg.output = {'brain' 'skull' 'scalp','tpm'}; > cfg.spmversion = 'spm8'; > cfg.name = 'test'; > cfg.write = 'yes'; > cfg.coordsys = 'spm'; > seg = ft_volumesegment(cfg,mri); > > Looking at the code for ft_volumesegment, it appears that a permutation of the mri dimensions (swapping 2 and 3) is done on line 279: > > [mri,permutevec,flipflags] = align_ijk2xyz(mri); > > The mri.transform field is changed accordingly. However, the original.transform is put into the headers for the c1, c2 and c3 output (line 399). I think that this is not correct because the segmentation output has the permuted dimensions, not the dimensions of the original mri. So, the correct transform should be the mri.transform after align_ijk2xyz I think. Or maybe this is intentional behavior and I'm missing something. It can easily be checked, however, using Check Reg, in SPM8, that the output tpm do not align correctly with the original mri. If I manually write the mri.transform (from align_ijk2xyz) to the tpm output, however, they do align correctly. > > Best, > > Don > > ----------------------- > Don Rojas, Ph.D. > Associate Professor of Psychiatry > U. of Colorado Denver Anschutz Medical Campus > Director, UCD Magnetoencephalography Lab > 13001 E. 17th Pl F546 > Aurora, CO 80045 > 303-724-4994 > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip Dr. J.M. (Jan-Mathijs) Schoffelen Donders Institute for Brain, Cognition and Behaviour, Centre for Cognitive Neuroimaging, Radboud University Nijmegen, The Netherlands J.Schoffelen at donders.ru.nl Telephone: +31-24-3614793 -------------- next part -------------- An HTML attachment was scrubbed... URL: From michael.wibral at web.de Mon Aug 8 11:22:18 2011 From: michael.wibral at web.de (Michael Wibral) Date: Mon, 8 Aug 2011 11:22:18 +0200 (CEST) Subject: [FieldTrip] Potential bug in trl Message-ID: <434769618.3170746.1312795338080.JavaMail.fmail@mwmweb077> An HTML attachment was scrubbed... URL: From e.maris at donders.ru.nl Mon Aug 8 11:23:39 2011 From: e.maris at donders.ru.nl (Eric Maris) Date: Mon, 8 Aug 2011 11:23:39 +0200 Subject: [FieldTrip] Extracting a timecourse from cluster of channels In-Reply-To: References: <4e34099a.c188e50a.3fc3.ffff8898SMTPIN_ADDED@mx.google.com> Message-ID: <071401cc55ac$dea9af20$9bfd0d60$@maris@donders.ru.nl> Hi Clara, > Do you think it's reasonable to take the intersection, or union, of > channels over time in a given space-time cluster? > Do you know of any publications that have dealt with this issue > (choosing channels from a space-time cluster for timecourse > extraction) that I could look at & reference? I wish I could, but I cannot help you here. Your question belongs to the category "important but not answerable on a principled basis (at least for me)". Best, Eric Maris > Thanks, Clara > > On Sat, Jul 30, 2011 at 9:31 AM, Eric Maris wrote: > > Dear Clara, > > > >> I am using fieldtrip's cluster-based permutation test to identify a > >> significant channel-time cluster showing an effect.  Now I would like > >> to extract a timecourse averaged over these channels (for the purpose > >> of comparing different conditions, not used to generate the cluster), > >> but I'm not sure if it makes sense to do so because the channels > >> belonging to the cluster change over time -- would the timecourse have > >> contributions from different channels at different time points?  Do I > >> need to limit my cluster to fixed channels across time? (And if so, > >> how do I do that?) > > > > Yes, you should fix channels across time. The most important point in > > selecting the channels is finding evidence for the fact that the channels > > reflects a single source only. This is not a statistical issue, and > > therefore there is not a p-value that can guide you in this choice. If you > > have collected MEG data, the best evidence is a dipolar topography of your > > effect. > > > > Best, > > > > Eric Maris > > > > > >> > >> Thanks! > >> Clara > >> _______________________________________________ > >> fieldtrip mailing list > >> fieldtrip at donders.ru.nl > >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > _______________________________________________ > > fieldtrip mailing list > > fieldtrip at donders.ru.nl > > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From drivolta81 at gmail.com Mon Aug 8 12:17:45 2011 From: drivolta81 at gmail.com (Davide Rivolta) Date: Mon, 8 Aug 2011 12:17:45 +0200 Subject: [FieldTrip] Potential bug in trl In-Reply-To: <434769618.3170746.1312795338080.JavaMail.fmail@mwmweb077> References: <434769618.3170746.1312795338080.JavaMail.fmail@mwmweb077> Message-ID: Dear Michael, After your initial reply I realised that I made a (silly) mistake in the trialfun. It was not a bug indeed. My apologies for this, Davide On Mon, Aug 8, 2011 at 11:22 AM, Michael Wibral wrote: > Hi Davide, > > would you mind sharing your solution (that could potentially help others as > well)? > > Thanks, > Michael > > > > ------------------------------ > *Von:* "Davide Rivolta" > *Gesendet:* Aug 5, 2011 11:36:07 AM > *An:* "Email discussion list for the FieldTrip project" < > fieldtrip at donders.ru.nl> > *Betreff:* Re: [FieldTrip] Potential bug in trl > > > Dear Jörn, > > Thank you for your reply. I guess I have now figured it out. > > > Thanks, > Davide > > On Fri, Aug 5, 2011 at 9:13 AM, "Jörn M. Horschig" < > jm.horschig at donders.ru.nl> wrote: > >> Dear Davide, >> >> I never worked with trials of different length, however this should be >> straight forward when writing your own trialfun (just as Michael proposed). >> In most trialfuns, the trl matrix is constructed by concatenating the >> vectors begsample, endsample and offset. These vectors can have any values >> you want them to have. As a simple example, you can write a trialfun that >> just returns trl = [1 2 1; 400 500 100] (for your experiment, you of course >> want these values to be computed from your trigger values). >> Is your problem that such a thing does not work? It would be great if you >> give a concrete use-case, or e.g. attach your trialfun. >> >> Best, >> Jörn >> >> >> >> On 8/4/2011 4:03 PM, Davide Rivolta wrote: >> >> Thanks Michael, >> >> I wrote my own trialfun. >> >> The offset would be negative and would have the same lenght of difference >> between endsample-begsample. This, for each trial, would have a different >> value. >> >> It can be a bug. In fact, I tried to vhange the trialfun by leaving the >> same begsample and endsample, but I put a fixed offset (i.e., -100) and it >> worked. All the trial was in and not, like with the variable offset, until >> time = 0. >> >> Thanks, >> >> Davide >> >> On Thu, Aug 4, 2011 at 3:55 PM, Michael Wibral wrote: >> >>> Hi davide, >>> >>> I guess in such a case you would have to write your own trial function. >>> However, offset values larger than 0 mean, that the stimulus occured before >>> you start your piece of data - is that really what you wanted ? >>> >>> Michael >>> >>> >>> ------------------------------ >>> *Von:* "Davide Rivolta" >>> *Gesendet:* Aug 4, 2011 2:44:08 PM >>> *An:* fieldtrip at donders.ru.nl >>> *Betreff:* [FieldTrip] Potential bug in trl >>> >>> >>> Dear all, >>> >>> I have noticed that the trial definition works well when begsample, >>> ensample and offset have fixed values. >>> >>> I have however a specific trialfunction, where each trial has a different >>> lenght. >>> >>> This works well when you specify a fixed offset, however, when even the >>> offset has to change for each trial, Fieldtrip reports values until 0. It >>> ignores time after 0. >>> >>> >>> Might it be a bug? >>> >>> Thanks, >>> >>> Davide >>> >>> >>> _______________________________________________ >>> fieldtrip mailing list >>> fieldtrip at donders.ru.nl >>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> >> >> >> >> -- >> Davide Rivolta, PhD >> >> >> >> _______________________________________________ >> fieldtrip mailing listfieldtrip at donders.ru.nlhttp://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> >> >> >> >> -- >> Jörn M. Horschig >> PhD Student >> Donders Institute for Brain, Cognition and Behaviour >> Centre for Cognitive Neuroimaging >> Radboud University Nijmegen >> Neuronal Oscillations Group >> >> P.O. Box 9101 >> NL-6500 HB Nijmegen >> The Netherlands >> >> Contact: >> E-Mail: jm.horschig at donders.ru.nl >> Tel: +31-(0)24-36-68493 >> Web: http://www.ru.nl/donders >> >> Visiting address: >> Trigon, room 2.30 >> Kapittelweg 29 >> NL-6525 EN Nijmegen >> The Netherlands >> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > > -- > Davide Rivolta, PhD > > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > -- Davide Rivolta, PhD -------------- next part -------------- An HTML attachment was scrubbed... URL: From marco.dahmane at gmail.com Mon Aug 8 15:13:24 2011 From: marco.dahmane at gmail.com (Marco Dahmane) Date: Mon, 8 Aug 2011 14:13:24 +0100 Subject: [FieldTrip] databrowser and merged data Message-ID: Hi all, I have a rather straightforward question. I would like to be able to visualize the output of ft_timelockanalysis in butterfly mode using the databrowser function. However, since my data was made by merging several raw datasets together, the sampleinfo field was erased during the process, and databrowser doesn't like that (even though the timelock.time and timelock.avg fields are fine) What I would like to know is, since I have quite a lot of small datasets to be merged together, is there a way to make FT automatically reconstruct the sampleinfo field after merging two datasets together ? In other words, do I *have* to manually add the sampleinfo field at the end (and how come it is lost upon merging two datasets?)? And if yes, what is the easiest way to go about it? Many thanks, --Marco -------------- next part -------------- An HTML attachment was scrubbed... URL: From eelke.spaak at donders.ru.nl Mon Aug 8 15:36:11 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Mon, 8 Aug 2011 15:36:11 +0200 Subject: [FieldTrip] databrowser and merged data In-Reply-To: References: Message-ID: Dear Marco, Let me make three points in response to your question. (1) Actually, in general, timelock data structures should never contain sampleinfo, since sampleinfo describes which samples in the original recording your trials correspond to (I'm not sure about timelocked data with cfg.keeptrials='yes'). (2) However, I don't think that point (1) is relevant to your problem. More relevant is this: why do you want to use the databrowser to visualise timelocked data? The databrowser is meant to conveniently page through data with a big time axis. Usually, your timelocked data will not have a big time axis, but a small one. For this, we have the ft_singleplotER-function, or simply matlab's plot(). (3) When you provide data that lacks a sampleinfo field to a fieldtrip function that requires a sampleinfo field, the sampleinfo should be automatically constructed. (This is taken care of by ft_checkdata, which is also responsible for converting your timelocked data into raw data in case you pass it to the databrowser.) So, even if you decide that you really want to use the databrowser to display timelocked data, it should actually just work :) Are you using the latest version of fieldtrip? If so, could you provide me with some more details on the error and its circumstances? Best, Eelke 2011/8/8 Marco Dahmane : > Hi all, > I have a rather straightforward question. I would like to be able to > visualize the output of ft_timelockanalysis in butterfly mode using the > databrowser function. > However, since my data was made by merging several raw datasets together, > the sampleinfo field was erased during the process, and databrowser doesn't > like that (even though the timelock.time and timelock.avg fields are fine) > What I would like to know is, since I have quite a lot of small datasets to > be merged together, is there a way to make FT automatically reconstruct the > sampleinfo field after merging two datasets together ? In other words, do I > *have* to manually add the sampleinfo field at the end (and how come it is > lost upon merging two datasets?)? > And if yes, what is the easiest way to go about it? > Many thanks, > --Marco > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From marco.dahmane at gmail.com Mon Aug 8 16:08:11 2011 From: marco.dahmane at gmail.com (Marco Dahmane) Date: Mon, 8 Aug 2011 15:08:11 +0100 Subject: [FieldTrip] databrowser and merged data Message-ID: Hi Eelke, So in order :) 1) Fair enough. And I'm not using cfg.keeptrials='yes' anyway 2) I'm actually using databrowser because I find it really convenient to just remove and add channels, or see to which channel each line corresponds. I know I could do this using Matlab's plot, but I was just wondering why it wouldn't work with databrowser. 3) Ok this is weird then. Something that will probably help you, whenever I use databrowser on my timelocked data, I get this : Warning: the trial definition in the configuration is inconsistent with the actual data > In utilities/private/warning_once at 75 In utilities/private/fixsampleinfo at 51 In ft_checkdata at 579 In ft_databrowser at 155 Warning: failed to create sampleinfo field > In utilities/private/warning_once at 75 In utilities/private/fixsampleinfo at 83 In ft_checkdata at 579 In ft_databrowser at 155 So there must be something wrong right? And yes I'm using the latest FT... Many thanks, --Marco -------------- next part -------------- An HTML attachment was scrubbed... URL: From BelluscB at ninds.nih.gov Mon Aug 8 18:08:16 2011 From: BelluscB at ninds.nih.gov (Belluscio, Beth (NIH/NINDS) [E]) Date: Mon, 8 Aug 2011 12:08:16 -0400 Subject: [FieldTrip] power line noise reduction Message-ID: <794DFDD3C128EF44AC49ADC58FD0090C057456B635@NIHMLBX10.nih.gov> Hi all- I am evaluating power using ft_freqanalysis for resting state data. In the preprocessing step, I indicate to remove power line noise at 60 Hz as follows: cfg = []; cfg.dataset = 'resting.ds'; cfg.channel = {'MEG'}; cfg.continuous = 'yes'; cfg.detrend = 'yes'; cfg.dftfilter = 'yes'; cfg.dftfreq = [60 120]; dataCB24rest = ft_preprocessing(cfg) Then I evaluate the power at frequencies from 5 to 100 Hz as follows: cfg = []; cfg.method = 'mtmfft'; cfg.taper = 'hanning'; cfg.foi = 5:5:100; cfg.tapsmofrq = 4; cfg.output = 'pow'; freqCB24rest = ft_freqanalysis(cfg, dataCB24rest) When I visualize the result, I get a VERY large value for the power at 60 Hz in every channel. For example, see attached. What am I doing wrong? Thanks for your help, Beth. Beth Belluscio MD-PhD Clinical Fellow Human Motor Control Section NINDS, NIH 301-402-3495 -------------- next part -------------- A non-text attachment was scrubbed... Name: example.tif Type: image/tiff Size: 18858 bytes Desc: example.tif URL: From jan.schoffelen at donders.ru.nl Mon Aug 8 19:43:40 2011 From: jan.schoffelen at donders.ru.nl (jan-mathijs schoffelen) Date: Mon, 8 Aug 2011 19:43:40 +0200 Subject: [FieldTrip] power line noise reduction In-Reply-To: <794DFDD3C128EF44AC49ADC58FD0090C057456B635@NIHMLBX10.nih.gov> References: <794DFDD3C128EF44AC49ADC58FD0090C057456B635@NIHMLBX10.nih.gov> Message-ID: <0E61E958-F55D-4D1C-B3CB-7F25C9E3A63E@donders.ru.nl> Hi Beth, It seems from your approach, that you use a long stretch of data on which you apply the dftfilter. The dftfilter only works well, if the power line fluctuations are stationary in amplitude. Another way of stating it, is that the line noise really has to be a very narrowband line in the spectrum (relative to the epoch length). Typically, the dftfilter works well only up to an epoch length of ~ 10 seconds (sometimes even less if there are sources of non-stationary power line interference). In your case I would either chop up the data in shorter segments, or use a notch filter (or a lowpassfilter; I would rather go for the low frequency stuff to begin with in the resting state). Best, Jan-Mathijs On Aug 8, 2011, at 6:08 PM, Belluscio, Beth (NIH/NINDS) [E] wrote: > Hi all- > I am evaluating power using ft_freqanalysis for resting state data. > In the preprocessing step, I indicate to remove power line noise at 60 Hz as follows: > > cfg = []; > cfg.dataset = 'resting.ds'; > cfg.channel = {'MEG'}; > cfg.continuous = 'yes'; > cfg.detrend = 'yes'; > cfg.dftfilter = 'yes'; > cfg.dftfreq = [60 120]; > dataCB24rest = ft_preprocessing(cfg) > > Then I evaluate the power at frequencies from 5 to 100 Hz as follows: > > cfg = []; > cfg.method = 'mtmfft'; > cfg.taper = 'hanning'; > cfg.foi = 5:5:100; > cfg.tapsmofrq = 4; > cfg.output = 'pow'; > freqCB24rest = ft_freqanalysis(cfg, dataCB24rest) > > When I visualize the result, I get a VERY large value for the power at 60 Hz in every channel. For example, see attached. > > What am I doing wrong? > Thanks for your help, > Beth. > > > Beth Belluscio MD-PhD > Clinical Fellow > Human Motor Control Section > NINDS, NIH > 301-402-3495 > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip Dr. J.M. (Jan-Mathijs) Schoffelen Donders Institute for Brain, Cognition and Behaviour, Centre for Cognitive Neuroimaging, Radboud University Nijmegen, The Netherlands J.Schoffelen at donders.ru.nl Telephone: +31-24-3614793 From ekanal at cmu.edu Mon Aug 8 20:07:26 2011 From: ekanal at cmu.edu (Kanal Eliezer) Date: Mon, 8 Aug 2011 14:07:26 -0400 Subject: [FieldTrip] question about re-doing analysis Message-ID: <150C06A5-3B96-4F2C-B529-DB1079A36701@cmu.edu> Hello folks - Simple question here... I've been working with a dataset for a while, and it turns out that the [1 50] bandpass filter I applied earlier is removing some data that we're going to want to examine. My question is, what's the best way to go about this? For most of the data, the preprocessing consisted of discarding trials (a lot of noisy trials, unfortunately) and ICA. I've saved the data from the preprocessing stages, and my current thought is to write a script to load up each dataset, extract the "sampleinfo" field, pull that from the original raw file, apply any new filters, and then work with that. This would mitigate my needing to go through all the data and do the trial reject again (which would take weeks of work). Is there a better way to do this? Thanks - Eliezer Kanal From jan.schoffelen at donders.ru.nl Mon Aug 8 20:33:48 2011 From: jan.schoffelen at donders.ru.nl (jan-mathijs schoffelen) Date: Mon, 8 Aug 2011 20:33:48 +0200 Subject: [FieldTrip] question about re-doing analysis In-Reply-To: <150C06A5-3B96-4F2C-B529-DB1079A36701@cmu.edu> References: <150C06A5-3B96-4F2C-B529-DB1079A36701@cmu.edu> Message-ID: <80BC061E-C9FD-46B5-B3D7-289260D35533@donders.ru.nl> Dear Elli, > to load up each dataset, extract the "sampleinfo" field, pull that from the original raw file, apply any new filters, and then work with that. This would mitigate my needing to go through all the data and do the trial reject again (which would take weeks of work). Is there a better way to do this? Isn't this good enough, then ;-) ? JM Dr. J.M. (Jan-Mathijs) Schoffelen Donders Institute for Brain, Cognition and Behaviour, Centre for Cognitive Neuroimaging, Radboud University Nijmegen, The Netherlands J.Schoffelen at donders.ru.nl Telephone: +31-24-3614793 From ekanal at cmu.edu Mon Aug 8 20:58:08 2011 From: ekanal at cmu.edu (Kanal Eliezer) Date: Mon, 8 Aug 2011 14:58:08 -0400 Subject: [FieldTrip] question about re-doing analysis In-Reply-To: <80BC061E-C9FD-46B5-B3D7-289260D35533@donders.ru.nl> References: <150C06A5-3B96-4F2C-B529-DB1079A36701@cmu.edu> <80BC061E-C9FD-46B5-B3D7-289260D35533@donders.ru.nl> Message-ID: <024FF5CF-ECE6-44ED-9C35-2D7636D797CF@cmu.edu> Yes, that works... the only problem with this approach is that I'll have to redo ICA filtering. I guess I was just curious if there's a different way that everyone uses that I'm not familiar with. This is the first dataset I've analyzed using fieldtrip, so I'm just trying to make sure I'm "doing it right". Thanks - Elli On Aug 8, 2011, at 2:33 PM, jan-mathijs schoffelen wrote: > Dear Elli, > >> to load up each dataset, extract the "sampleinfo" field, pull that from the original raw file, apply any new filters, and then work with that. This would mitigate my needing to go through all the data and do the trial reject again (which would take weeks of work). Is there a better way to do this? > > Isn't this good enough, then ;-) ? > > JM > > > > Dr. J.M. (Jan-Mathijs) Schoffelen > Donders Institute for Brain, Cognition and Behaviour, > Centre for Cognitive Neuroimaging, > Radboud University Nijmegen, The Netherlands > J.Schoffelen at donders.ru.nl > Telephone: +31-24-3614793 > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From jan.schoffelen at donders.ru.nl Mon Aug 8 21:05:12 2011 From: jan.schoffelen at donders.ru.nl (jan-mathijs schoffelen) Date: Mon, 8 Aug 2011 21:05:12 +0200 Subject: [FieldTrip] question about re-doing analysis In-Reply-To: <024FF5CF-ECE6-44ED-9C35-2D7636D797CF@cmu.edu> References: <150C06A5-3B96-4F2C-B529-DB1079A36701@cmu.edu> <80BC061E-C9FD-46B5-B3D7-289260D35533@donders.ru.nl> <024FF5CF-ECE6-44ED-9C35-2D7636D797CF@cmu.edu> Message-ID: <6044F712-0B26-4637-B05B-9E89E9A7D206@donders.ru.nl> Hi Elli, Yes, that's a good point. What I sometimes do, is to also store the mixing matrix which comes out after ft_componentanalysis (i.e. the comp.topo). When you have this matrix, you can very cheaply re-ICA-filter the data. Yet, note that with different filter settings in the preprocessing, the ICA-components (and topographies) may also change. BW, JM On Aug 8, 2011, at 8:58 PM, Kanal Eliezer wrote: > Yes, that works... the only problem with this approach is that I'll have to redo ICA filtering. > > I guess I was just curious if there's a different way that everyone uses that I'm not familiar with. This is the first dataset I've analyzed using fieldtrip, so I'm just trying to make sure I'm "doing it right". Thanks - > > Elli > > > On Aug 8, 2011, at 2:33 PM, jan-mathijs schoffelen wrote: > >> Dear Elli, >> >>> to load up each dataset, extract the "sampleinfo" field, pull that from the original raw file, apply any new filters, and then work with that. This would mitigate my needing to go through all the data and do the trial reject again (which would take weeks of work). Is there a better way to do this? >> >> Isn't this good enough, then ;-) ? >> >> JM >> >> >> >> Dr. J.M. (Jan-Mathijs) Schoffelen >> Donders Institute for Brain, Cognition and Behaviour, >> Centre for Cognitive Neuroimaging, >> Radboud University Nijmegen, The Netherlands >> J.Schoffelen at donders.ru.nl >> Telephone: +31-24-3614793 >> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip Dr. J.M. (Jan-Mathijs) Schoffelen Donders Institute for Brain, Cognition and Behaviour, Centre for Cognitive Neuroimaging, Radboud University Nijmegen, The Netherlands J.Schoffelen at donders.ru.nl Telephone: +31-24-3614793 From ekanal at cmu.edu Mon Aug 8 21:19:42 2011 From: ekanal at cmu.edu (Kanal Eliezer) Date: Mon, 8 Aug 2011 15:19:42 -0400 Subject: [FieldTrip] question about re-doing analysis In-Reply-To: <6044F712-0B26-4637-B05B-9E89E9A7D206@donders.ru.nl> References: <150C06A5-3B96-4F2C-B529-DB1079A36701@cmu.edu> <80BC061E-C9FD-46B5-B3D7-289260D35533@donders.ru.nl> <024FF5CF-ECE6-44ED-9C35-2D7636D797CF@cmu.edu> <6044F712-0B26-4637-B05B-9E89E9A7D206@donders.ru.nl> Message-ID: <5B688DA3-CECE-42F5-9E0C-70F3FCB086FB@cmu.edu> That's a good point, I guess I'll have needed to re-do them anyway. Elli On Aug 8, 2011, at 3:05 PM, jan-mathijs schoffelen wrote: > Hi Elli, > > Yes, that's a good point. What I sometimes do, is to also store the mixing matrix which comes out after ft_componentanalysis (i.e. the comp.topo). When you have this matrix, you can very cheaply re-ICA-filter the data. Yet, note that with different filter settings in the preprocessing, the ICA-components (and topographies) may also change. > > > BW, JM > > > > On Aug 8, 2011, at 8:58 PM, Kanal Eliezer wrote: > >> Yes, that works... the only problem with this approach is that I'll have to redo ICA filtering. >> >> I guess I was just curious if there's a different way that everyone uses that I'm not familiar with. This is the first dataset I've analyzed using fieldtrip, so I'm just trying to make sure I'm "doing it right". Thanks - >> >> Elli >> >> >> On Aug 8, 2011, at 2:33 PM, jan-mathijs schoffelen wrote: >> >>> Dear Elli, >>> >>>> to load up each dataset, extract the "sampleinfo" field, pull that from the original raw file, apply any new filters, and then work with that. This would mitigate my needing to go through all the data and do the trial reject again (which would take weeks of work). Is there a better way to do this? >>> >>> Isn't this good enough, then ;-) ? >>> >>> JM >>> >>> >>> >>> Dr. J.M. (Jan-Mathijs) Schoffelen >>> Donders Institute for Brain, Cognition and Behaviour, >>> Centre for Cognitive Neuroimaging, >>> Radboud University Nijmegen, The Netherlands >>> J.Schoffelen at donders.ru.nl >>> Telephone: +31-24-3614793 >>> >>> >>> _______________________________________________ >>> fieldtrip mailing list >>> fieldtrip at donders.ru.nl >>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > Dr. J.M. (Jan-Mathijs) Schoffelen > Donders Institute for Brain, Cognition and Behaviour, > Centre for Cognitive Neuroimaging, > Radboud University Nijmegen, The Netherlands > J.Schoffelen at donders.ru.nl > Telephone: +31-24-3614793 > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From eelke.spaak at donders.ru.nl Tue Aug 9 11:00:27 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Tue, 9 Aug 2011 11:00:27 +0200 Subject: [FieldTrip] databrowser and merged data In-Reply-To: References: Message-ID: Hi Marco, The subfunction fixsampleinfo (used by ft_checkdata) attempts a number of things in order to create a sampleinfo field. My guess is that it detected a trial definition somewhere in your cfg "tree" (the data.cfg and data.cfg.previous fields) that was inconsistent with your data, and therefore refused to continue rebuilding a sampleinfo from scratch. I have now changed this function so that it will always reconstruct your sampleinfo, even if it detects an inconsistent trial definition in a previous configuration structure. So, it should now work :) At least, with tomorrow's update on the FTP server. Best, Eelke 2011/8/8 Marco Dahmane : > Hi Eelke, > So in order :) > 1) Fair enough. And I'm not using cfg.keeptrials='yes' anyway > 2) I'm actually using databrowser because I find it really convenient to > just remove and add channels, or see to which channel each line corresponds. > I know I could do this using Matlab's plot, but I was just wondering why it > wouldn't work with databrowser. > 3) Ok this is weird then. Something that will probably help you, whenever I > use databrowser on my timelocked data, I get this : > Warning: the trial definition in the configuration is inconsistent with the > actual data >> In utilities/private/warning_once at 75 >   In utilities/private/fixsampleinfo at 51 >   In ft_checkdata at 579 >   In ft_databrowser at 155 > Warning: failed to create sampleinfo field >> In utilities/private/warning_once at 75 >   In utilities/private/fixsampleinfo at 83 >   In ft_checkdata at 579 >   In ft_databrowser at 155 >  So there must be something wrong right? > And yes I'm using the latest FT... > Many thanks, > --Marco > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From jm.horschig at donders.ru.nl Tue Aug 9 12:21:45 2011 From: jm.horschig at donders.ru.nl (=?ISO-8859-1?Q?=22J=F6rn_M=2E_Horschig=22?=) Date: Tue, 09 Aug 2011 12:21:45 +0200 Subject: [FieldTrip] databrowser and merged data In-Reply-To: References: Message-ID: <4E410A39.2060809@donders.ru.nl> Hi Marco and Eelke, sampleinfo cannot be reconstructed after resampling the data, because there is no way to precisely relate your resampled samples to the original samples. My guess would be that you, Marco, did that and therefore encounter the problem you mentioned. If that was so, I am not sure whether fixsampleinfo should actually recompute the sampleinfo field, because it makes absolutely no sense to have a sampleinfo field in your resampled data that relates back samples in the original data. Anyway, if you did not resample your data, then fixsampleinfo should work. Could you let us know how you preprocessed your data? Best, Jörn On 8/9/2011 11:00 AM, Eelke Spaak wrote: > Hi Marco, > > The subfunction fixsampleinfo (used by ft_checkdata) attempts a number > of things in order to create a sampleinfo field. My guess is that it > detected a trial definition somewhere in your cfg "tree" (the data.cfg > and data.cfg.previous fields) that was inconsistent with your data, > and therefore refused to continue rebuilding a sampleinfo from > scratch. I have now changed this function so that it will always > reconstruct your sampleinfo, even if it detects an inconsistent trial > definition in a previous configuration structure. > > So, it should now work :) At least, with tomorrow's update on the FTP server. > > Best, > Eelke > > 2011/8/8 Marco Dahmane: >> Hi Eelke, >> So in order :) >> 1) Fair enough. And I'm not using cfg.keeptrials='yes' anyway >> 2) I'm actually using databrowser because I find it really convenient to >> just remove and add channels, or see to which channel each line corresponds. >> I know I could do this using Matlab's plot, but I was just wondering why it >> wouldn't work with databrowser. >> 3) Ok this is weird then. Something that will probably help you, whenever I >> use databrowser on my timelocked data, I get this : >> Warning: the trial definition in the configuration is inconsistent with the >> actual data >>> In utilities/private/warning_once at 75 >> In utilities/private/fixsampleinfo at 51 >> In ft_checkdata at 579 >> In ft_databrowser at 155 >> Warning: failed to create sampleinfo field >>> In utilities/private/warning_once at 75 >> In utilities/private/fixsampleinfo at 83 >> In ft_checkdata at 579 >> In ft_databrowser at 155 >> So there must be something wrong right? >> And yes I'm using the latest FT... >> Many thanks, >> --Marco >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip -- Jörn M. Horschig PhD Student Donders Institute for Brain, Cognition and Behaviour Centre for Cognitive Neuroimaging Radboud University Nijmegen Neuronal Oscillations Group P.O. Box 9101 NL-6500 HB Nijmegen The Netherlands Contact: E-Mail: jm.horschig at donders.ru.nl Tel: +31-(0)24-36-68493 Web: http://www.ru.nl/donders Visiting address: Trigon, room 2.30 Kapittelweg 29 NL-6525 EN Nijmegen The Netherlands From eelke.spaak at donders.ru.nl Tue Aug 9 12:28:21 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Tue, 9 Aug 2011 12:28:21 +0200 Subject: [FieldTrip] databrowser and merged data In-Reply-To: <4E410A39.2060809@donders.ru.nl> References: <4E410A39.2060809@donders.ru.nl> Message-ID: Hi Jörn, As you can read in my email, I actually changed fixsampleinfo so that it *does* reconstruct the sample info now :) I agree that sampleinfo does not make sense in the case of data that is (1) timelock-averaged and (2) combined from multiple datasets, but the issue here was to get the databrowser working on such data. Since the data used internally by the databrowser (after the call to ft_checkdata) is never exposed to the user, this seems like a good solution to me. Best, Eelke 2011/8/9 "Jörn M. Horschig" : > Hi Marco and Eelke, > > sampleinfo cannot be reconstructed after resampling the data, because there > is no way to precisely relate your resampled samples to the original > samples. My guess would be that you, Marco, did that and therefore encounter > the problem you mentioned. If that was so, I am not sure whether > fixsampleinfo should actually recompute the sampleinfo field, because it > makes absolutely no sense to have a sampleinfo field in your resampled data > that relates back samples in the original data. > Anyway, if you did not resample your data, then fixsampleinfo should work. > Could you let us know how you preprocessed your data? > > Best, > Jörn > > On 8/9/2011 11:00 AM, Eelke Spaak wrote: >> >> Hi Marco, >> >> The subfunction fixsampleinfo (used by ft_checkdata) attempts a number >> of things in order to create a sampleinfo field. My guess is that it >> detected a trial definition somewhere in your cfg "tree" (the data.cfg >> and data.cfg.previous fields) that was inconsistent with your data, >> and therefore refused to continue rebuilding a sampleinfo from >> scratch. I have now changed this function so that it will always >> reconstruct your sampleinfo, even if it detects an inconsistent trial >> definition in a previous configuration structure. >> >> So, it should now work :) At least, with tomorrow's update on the FTP >> server. >> >> Best, >> Eelke >> >> 2011/8/8 Marco Dahmane: >>> >>> Hi Eelke, >>> So in order :) >>> 1) Fair enough. And I'm not using cfg.keeptrials='yes' anyway >>> 2) I'm actually using databrowser because I find it really convenient to >>> just remove and add channels, or see to which channel each line >>> corresponds. >>> I know I could do this using Matlab's plot, but I was just wondering why >>> it >>> wouldn't work with databrowser. >>> 3) Ok this is weird then. Something that will probably help you, whenever >>> I >>> use databrowser on my timelocked data, I get this : >>> Warning: the trial definition in the configuration is inconsistent with >>> the >>> actual data >>>> >>>> In utilities/private/warning_once at 75 >>> >>>   In utilities/private/fixsampleinfo at 51 >>>   In ft_checkdata at 579 >>>   In ft_databrowser at 155 >>> Warning: failed to create sampleinfo field >>>> >>>> In utilities/private/warning_once at 75 >>> >>>   In utilities/private/fixsampleinfo at 83 >>>   In ft_checkdata at 579 >>>   In ft_databrowser at 155 >>>  So there must be something wrong right? >>> And yes I'm using the latest FT... >>> Many thanks, >>> --Marco >>> _______________________________________________ >>> fieldtrip mailing list >>> fieldtrip at donders.ru.nl >>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >>> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > -- > Jörn M. Horschig > PhD Student > Donders Institute for Brain, Cognition and Behaviour > Centre for Cognitive Neuroimaging > Radboud University Nijmegen > Neuronal Oscillations Group > > P.O. Box 9101 > NL-6500 HB Nijmegen > The Netherlands > > Contact: > E-Mail: jm.horschig at donders.ru.nl > Tel:    +31-(0)24-36-68493 > Web: http://www.ru.nl/donders > > Visiting address: > Trigon, room 2.30 > Kapittelweg 29 > NL-6525 EN Nijmegen > The Netherlands > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From eelke.spaak at donders.ru.nl Tue Aug 9 12:30:07 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Tue, 9 Aug 2011 12:30:07 +0200 Subject: [FieldTrip] databrowser and merged data In-Reply-To: References: <4E410A39.2060809@donders.ru.nl> Message-ID: PS: In other words, fixsampleinfo now generates an ad-hoc sampleinfo with no relation to the original dataset, if requested to do so. Marco's preprocessing steps are not relevant for this. 2011/8/9 Eelke Spaak : > Hi Jörn, > > As you can read in my email, I actually changed fixsampleinfo so that > it *does* reconstruct the sample info now :) > > I agree that sampleinfo does not make sense in the case of data that > is (1) timelock-averaged and (2) combined from multiple datasets, but > the issue here was to get the databrowser working on such data. Since > the data used internally by the databrowser (after the call to > ft_checkdata) is never exposed to the user, this seems like a good > solution to me. > > Best, > Eelke > > 2011/8/9 "Jörn M. Horschig" : >> Hi Marco and Eelke, >> >> sampleinfo cannot be reconstructed after resampling the data, because there >> is no way to precisely relate your resampled samples to the original >> samples. My guess would be that you, Marco, did that and therefore encounter >> the problem you mentioned. If that was so, I am not sure whether >> fixsampleinfo should actually recompute the sampleinfo field, because it >> makes absolutely no sense to have a sampleinfo field in your resampled data >> that relates back samples in the original data. >> Anyway, if you did not resample your data, then fixsampleinfo should work. >> Could you let us know how you preprocessed your data? >> >> Best, >> Jörn >> >> On 8/9/2011 11:00 AM, Eelke Spaak wrote: >>> >>> Hi Marco, >>> >>> The subfunction fixsampleinfo (used by ft_checkdata) attempts a number >>> of things in order to create a sampleinfo field. My guess is that it >>> detected a trial definition somewhere in your cfg "tree" (the data.cfg >>> and data.cfg.previous fields) that was inconsistent with your data, >>> and therefore refused to continue rebuilding a sampleinfo from >>> scratch. I have now changed this function so that it will always >>> reconstruct your sampleinfo, even if it detects an inconsistent trial >>> definition in a previous configuration structure. >>> >>> So, it should now work :) At least, with tomorrow's update on the FTP >>> server. >>> >>> Best, >>> Eelke >>> >>> 2011/8/8 Marco Dahmane: >>>> >>>> Hi Eelke, >>>> So in order :) >>>> 1) Fair enough. And I'm not using cfg.keeptrials='yes' anyway >>>> 2) I'm actually using databrowser because I find it really convenient to >>>> just remove and add channels, or see to which channel each line >>>> corresponds. >>>> I know I could do this using Matlab's plot, but I was just wondering why >>>> it >>>> wouldn't work with databrowser. >>>> 3) Ok this is weird then. Something that will probably help you, whenever >>>> I >>>> use databrowser on my timelocked data, I get this : >>>> Warning: the trial definition in the configuration is inconsistent with >>>> the >>>> actual data >>>>> >>>>> In utilities/private/warning_once at 75 >>>> >>>>   In utilities/private/fixsampleinfo at 51 >>>>   In ft_checkdata at 579 >>>>   In ft_databrowser at 155 >>>> Warning: failed to create sampleinfo field >>>>> >>>>> In utilities/private/warning_once at 75 >>>> >>>>   In utilities/private/fixsampleinfo at 83 >>>>   In ft_checkdata at 579 >>>>   In ft_databrowser at 155 >>>>  So there must be something wrong right? >>>> And yes I'm using the latest FT... >>>> Many thanks, >>>> --Marco >>>> _______________________________________________ >>>> fieldtrip mailing list >>>> fieldtrip at donders.ru.nl >>>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >>>> >>> _______________________________________________ >>> fieldtrip mailing list >>> fieldtrip at donders.ru.nl >>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> >> >> -- >> Jörn M. Horschig >> PhD Student >> Donders Institute for Brain, Cognition and Behaviour >> Centre for Cognitive Neuroimaging >> Radboud University Nijmegen >> Neuronal Oscillations Group >> >> P.O. Box 9101 >> NL-6500 HB Nijmegen >> The Netherlands >> >> Contact: >> E-Mail: jm.horschig at donders.ru.nl >> Tel:    +31-(0)24-36-68493 >> Web: http://www.ru.nl/donders >> >> Visiting address: >> Trigon, room 2.30 >> Kapittelweg 29 >> NL-6525 EN Nijmegen >> The Netherlands >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> > From jm.horschig at donders.ru.nl Tue Aug 9 14:11:34 2011 From: jm.horschig at donders.ru.nl (=?ISO-8859-1?Q?=22J=F6rn_M=2E_Horschig=22?=) Date: Tue, 09 Aug 2011 14:11:34 +0200 Subject: [FieldTrip] databrowser and merged data In-Reply-To: References: <4E410A39.2060809@donders.ru.nl> Message-ID: <4E4123F6.9000209@donders.ru.nl> Hi Eelke, I have read your mail and would like to emphasize that I disagree. If a sampleinfo field is constructed that cannot be used to infer anything from, it should not be done. Thus, the fix should not be in fixsampleinfo, because it will result in all kind of other issues when it comes to reconstructing sampleinfo, although it might help to solve this particular problem. Let's discuss this at the next FT meeting rather than here :) Best, Jörn On 8/9/2011 12:30 PM, Eelke Spaak wrote: > PS: In other words, fixsampleinfo now generates an ad-hoc sampleinfo > with no relation to the original dataset, if requested to do so. > Marco's preprocessing steps are not relevant for this. > > 2011/8/9 Eelke Spaak: >> Hi Jörn, >> >> As you can read in my email, I actually changed fixsampleinfo so that >> it *does* reconstruct the sample info now :) >> >> I agree that sampleinfo does not make sense in the case of data that >> is (1) timelock-averaged and (2) combined from multiple datasets, but >> the issue here was to get the databrowser working on such data. Since >> the data used internally by the databrowser (after the call to >> ft_checkdata) is never exposed to the user, this seems like a good >> solution to me. >> >> Best, >> Eelke >> >> 2011/8/9 "Jörn M. Horschig": >>> Hi Marco and Eelke, >>> >>> sampleinfo cannot be reconstructed after resampling the data, because there >>> is no way to precisely relate your resampled samples to the original >>> samples. My guess would be that you, Marco, did that and therefore encounter >>> the problem you mentioned. If that was so, I am not sure whether >>> fixsampleinfo should actually recompute the sampleinfo field, because it >>> makes absolutely no sense to have a sampleinfo field in your resampled data >>> that relates back samples in the original data. >>> Anyway, if you did not resample your data, then fixsampleinfo should work. >>> Could you let us know how you preprocessed your data? >>> >>> Best, >>> Jörn >>> >>> On 8/9/2011 11:00 AM, Eelke Spaak wrote: >>>> Hi Marco, >>>> >>>> The subfunction fixsampleinfo (used by ft_checkdata) attempts a number >>>> of things in order to create a sampleinfo field. My guess is that it >>>> detected a trial definition somewhere in your cfg "tree" (the data.cfg >>>> and data.cfg.previous fields) that was inconsistent with your data, >>>> and therefore refused to continue rebuilding a sampleinfo from >>>> scratch. I have now changed this function so that it will always >>>> reconstruct your sampleinfo, even if it detects an inconsistent trial >>>> definition in a previous configuration structure. >>>> >>>> So, it should now work :) At least, with tomorrow's update on the FTP >>>> server. >>>> >>>> Best, >>>> Eelke >>>> >>>> 2011/8/8 Marco Dahmane: >>>>> Hi Eelke, >>>>> So in order :) >>>>> 1) Fair enough. And I'm not using cfg.keeptrials='yes' anyway >>>>> 2) I'm actually using databrowser because I find it really convenient to >>>>> just remove and add channels, or see to which channel each line >>>>> corresponds. >>>>> I know I could do this using Matlab's plot, but I was just wondering why >>>>> it >>>>> wouldn't work with databrowser. >>>>> 3) Ok this is weird then. Something that will probably help you, whenever >>>>> I >>>>> use databrowser on my timelocked data, I get this : >>>>> Warning: the trial definition in the configuration is inconsistent with >>>>> the >>>>> actual data >>>>>> In utilities/private/warning_once at 75 >>>>> In utilities/private/fixsampleinfo at 51 >>>>> In ft_checkdata at 579 >>>>> In ft_databrowser at 155 >>>>> Warning: failed to create sampleinfo field >>>>>> In utilities/private/warning_once at 75 >>>>> In utilities/private/fixsampleinfo at 83 >>>>> In ft_checkdata at 579 >>>>> In ft_databrowser at 155 >>>>> So there must be something wrong right? >>>>> And yes I'm using the latest FT... >>>>> Many thanks, >>>>> --Marco >>>>> _______________________________________________ >>>>> fieldtrip mailing list >>>>> fieldtrip at donders.ru.nl >>>>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >>>>> >>>> _______________________________________________ >>>> fieldtrip mailing list >>>> fieldtrip at donders.ru.nl >>>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >>> >>> -- >>> Jörn M. Horschig >>> PhD Student >>> Donders Institute for Brain, Cognition and Behaviour >>> Centre for Cognitive Neuroimaging >>> Radboud University Nijmegen >>> Neuronal Oscillations Group >>> >>> P.O. Box 9101 >>> NL-6500 HB Nijmegen >>> The Netherlands >>> >>> Contact: >>> E-Mail: jm.horschig at donders.ru.nl >>> Tel: +31-(0)24-36-68493 >>> Web: http://www.ru.nl/donders >>> >>> Visiting address: >>> Trigon, room 2.30 >>> Kapittelweg 29 >>> NL-6525 EN Nijmegen >>> The Netherlands >>> >>> _______________________________________________ >>> fieldtrip mailing list >>> fieldtrip at donders.ru.nl >>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >>> > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip -- Jörn M. Horschig PhD Student Donders Institute for Brain, Cognition and Behaviour Centre for Cognitive Neuroimaging Radboud University Nijmegen Neuronal Oscillations Group P.O. Box 9101 NL-6500 HB Nijmegen The Netherlands Contact: E-Mail: jm.horschig at donders.ru.nl Tel: +31-(0)24-36-68493 Web: http://www.ru.nl/donders Visiting address: Trigon, room 2.30 Kapittelweg 29 NL-6525 EN Nijmegen The Netherlands From hubert.preissl at uni-tuebingen.de Wed Aug 10 11:16:08 2011 From: hubert.preissl at uni-tuebingen.de (Hubert Preissl) Date: Wed, 10 Aug 2011 11:16:08 +0200 Subject: [FieldTrip] =?windows-1252?q?Autumn_School_=93The_multimodal_brai?= =?windows-1252?q?n=94=3A_5th-6th_October_2011=2C_T=FCbingen?= In-Reply-To: <5B688DA3-CECE-42F5-9E0C-70F3FCB086FB@cmu.edu> References: <150C06A5-3B96-4F2C-B529-DB1079A36701@cmu.edu> <80BC061E-C9FD-46B5-B3D7-289260D35533@donders.ru.nl> <024FF5CF-ECE6-44ED-9C35-2D7636D797CF@cmu.edu> <6044F712-0B26-4637-B05B-9E89E9A7D206@donders.ru.nl> <5B688DA3-CECE-42F5-9E0C-70F3FCB086FB@cmu.edu> Message-ID: <4E424C58.6000406@uni-tuebingen.de> *Autumn School “The multimodal brain”: 5^th -6^th October 2011* Hello, we are pleased to announce the upcoming “Autumn school” in Tübingen organized by the MEG Center. For application and further information on invited speakers and scientific program please visit our website: http://www.mp.uni-tuebingen.de/mp/index.php?id=396 We look forward to meet you in Tübingen! Best regards , Hubert Preissl ps. sorry if you receive this mail several times based on some cross-posting. -- Dr. Hubert Preissl MEG Center phone: ++49-(0)7071-2987704 Otfried Müller Str 47 fax: ++49-(0)7071-295706 72076 Tübingen, Germany email: hubert.preissl at uni-tuebingen.de -------------- next part -------------- An HTML attachment was scrubbed... URL: From sylvana.schister at utah.edu Fri Aug 12 00:56:44 2011 From: sylvana.schister at utah.edu (Sylvana Schister) Date: Thu, 11 Aug 2011 16:56:44 -0600 Subject: [FieldTrip] Alignment of MRI and estimated sources Message-ID: <1313103404.26076.16.camel@sylvana-desktop> Hello Fieldtrippers! I am having trouble getting meaningful results after performing source analysis with DICS due to a misalignment of my MRI dataset. I am relatively new to FT and I will really appreciate your help. I am using the 'standard_BEM' files provided in the tutorials. I manage to realign the MRI to head coordinates using volumerealign, reslice and segment it. Then, I use the modified MRI to compute the forward model. The problem arises when proceeding to ft_sourceinterpolate and ft_sourceanalysis. As I understand, the 'mri' input to ft_sourceinterpolate must be the output of FT_READ_MRI or the filename of a MRI. The function crashes when using my realigned version of the data, therefore I am using the original file, 'standard_mri.mat'. This file contains no fiducials. No surprise the results are then messed up. How/ where do I need to specify the fiducials? Why can't I use the aligned mri? What is the correct way to do this? Thank you so much for your help! Sylvana From Santosh.VemaKrishnamurthy at nrc-cnrc.gc.ca Fri Aug 12 16:00:10 2011 From: Santosh.VemaKrishnamurthy at nrc-cnrc.gc.ca (Vema Krishnamurthy, Santosh) Date: Fri, 12 Aug 2011 07:00:10 -0700 Subject: [FieldTrip] Scalp segmentation from an MRI (.nii or .fif) Message-ID: Hello Everyone, I'm very new to Fieldtrip and I need to perform scalp segmentation from an mri file(.nii or .fif). I'm working on a project where I need the scalp surface for surface matching, so I need the scalp surface with high resolution which clearly shows the facial features (nose, eyes, lips, chin). Any help/suggestions relating to this are much appreciated. Thanks and Regards, --------------------------------------------------- Santosh Vema KrishnaMurthy, M.A.Sc Medical Devices Engineer NRC Institute for Biodiagnostics (Atlantic) Elekta/IWK MEG Research office Ph: (902) 470-3984 --------------------------------------------------- -------------- next part -------------- An HTML attachment was scrubbed... URL: From Gregor.Volberg at psychologie.uni-regensburg.de Fri Aug 12 18:47:30 2011 From: Gregor.Volberg at psychologie.uni-regensburg.de (Gregor Volberg) Date: Fri, 12 Aug 2011 18:47:30 +0200 Subject: [FieldTrip] Antw: Scalp segmentation from an MRI (.nii or .fif) Message-ID: <4E457543020000570000A631@gwsmtp1.uni-regensburg.de> Dear Santosh, for that purpose you might want to have a look at the NFT toolbox from SCCN, http://sccn.ucsd.edu/nft/ . It does high resolution segmentations for brain, csf, scalp and skin surfaces. If you need the segmentation for computing forward solutions with BEM head models, then I would discourage a too fine resolution, though. It strongly increases the computational load. Best regards, Gregor -- Dr. rer. nat. Gregor Volberg ( mailto:gregor.volberg at psychologie.uni-regensburg.de ) University of Regensburg Institute for Experimental Psychology 93040 Regensburg, Germany Tel: +49 941 943 3862 Fax: +49 941 943 3233 http://www.psychologie.uni-regensburg.de/Greenlee/team/volberg/volberg.html >>> "Vema Krishnamurthy, Santosh" 12.08.11 16.42 Uhr >>> Hello Everyone, I'm very new to Fieldtrip and I need to perform scalp segmentation from an mri file(.nii or .fif). I'm working on a project where I need the scalp surface for surface matching, so I need the scalp surface with high resolution which clearly shows the facial features (nose, eyes, lips, chin). Any help/suggestions relating to this are much appreciated. Thanks and Regards, --------------------------------------------------- Santosh Vema KrishnaMurthy, M.A.Sc Medical Devices Engineer NRC Institute for Biodiagnostics (Atlantic) Elekta/IWK MEG Research office Ph: (902) 470-3984 --------------------------------------------------- -------------- next part -------------- An HTML attachment was scrubbed... URL: From jan.schoffelen at donders.ru.nl Sat Aug 13 18:23:49 2011 From: jan.schoffelen at donders.ru.nl (jan-mathijs schoffelen) Date: Sat, 13 Aug 2011 18:23:49 +0200 Subject: [FieldTrip] Fwd: ft_definetrial crash References: Message-ID: <4B01B4FC-FF35-435F-8167-F2A82C0D1B9E@donders.ru.nl> Hi Tobias, I forward your question to the discussion list, so that many may benefit from the discussion. I am not sure what is going on and I don't know whether other people have encountered this. Are you calling the function from the correct directory? In other words, you should not be within the *.ds directory itself. You could also try to explicitly define cfg.datafile and cfg.headerfile (and omit the cfg.dataset). Let me know when the problem persists. Best wishes, Jan-Mathijs Begin forwarded message: > From: Tobias Overath > Date: August 3, 2011 12:37:15 PM GMT+02:00 > To: jan.schoffelen at donders.ru.nl > Subject: ft_definetrial crash > > Hi Jan-Mathijs, > > I participated in the MEG FieldTrip course this past April and have finally managed to get my first MEG dataset from the CTF machine here at the FIL. However, I am already crashing at the first step, i.e. when running ft_definetrial. > > This is my code: > cfg = []; > cfg.dataset = 'S01_01.ds'; > cfg.trialdef.eventtype = 'frontpanel trigger'; > cfg.trialdef.eventvalue = 1; > cfg.trialdef.prestim = .25; > cfg.trialdef.poststim = 2.75; > cfg = ft_definetrial(cfg); > > It recognises correctly that it's supposed to read the header from S01_01.ds/S01_01.res, but it then tells me that it can't find the file S01_01.meg4 (which is true, because it should be S01_01.ds/S01_01.meg4). I presume this is a pretty common error, but it occurs with two FieldTrip versions that I have downloaded (very recent ones). I am probably missing something very obvious... > > Incidentally, is there no other way to search the FieldTrip email archives than to download a zipped text file for each month of the previous 7 years?!? > > Thanks, > > Tobias > > > > -- > Tobias Overath, PhD > UCL Ear Institute > University College London > 332 Grays Inn Road > London, WC1X 8EE > UK > http://www.homepages.ucl.ac.uk/~skgtjto > > > > > Dr. J.M. (Jan-Mathijs) Schoffelen Donders Institute for Brain, Cognition and Behaviour, Centre for Cognitive Neuroimaging, Radboud University Nijmegen, The Netherlands J.Schoffelen at donders.ru.nl Telephone: +31-24-3614793 -------------- next part -------------- An HTML attachment was scrubbed... URL: From r.oostenveld at donders.ru.nl Mon Aug 15 10:10:06 2011 From: r.oostenveld at donders.ru.nl (Robert Oostenveld) Date: Mon, 15 Aug 2011 10:10:06 +0200 Subject: [FieldTrip] Invitation to contribute to a Special Issue on Brain Oscillations during Language Processing References: Message-ID: <021E015B-2CB1-46FA-96E3-0DA3922CEB58@donders.ru.nl> Begin forwarded message: > Dear colleagues, > We would like to invite you to contribute your research to our special issue on the role of brain oscillations in language processing, to appear in Frontiers in Language Science. > You can visit the web site at: > http://www.frontiersin.org/languagesciences/specialtopics/brain_oscillations_during_lang/397 > See detailed description below. > The call has been very successful so far and prominent figures in the field have joined us in this project. > We are looking forward to receiving your research. > Best wishes, > Lucia Melloni & Marcela Pena > > > Brain Oscillations during Language Processing: from Perception to Production > > Deadline for abstract submission: 01 Sep 2011 > Deadline for full article submission: 15 Dec 2011 > > Language processing is a seemingly effortless task that requires the integration of speech units (e.g., phonemes, syllables, words, etc.) occurring at different rates. In particular, temporal binding for speech should occur within and across different temporal scales, necessitating multiple simultaneous windows of integration for prosodic, semantic, syntactic and pragmatic processing. Recent evidence suggests that neuronal oscillations may reflect both tracking linguistic units at their individual rhythms as well as integrating speech units over a large range of temporal scales. > > The present Research Topic would like to evaluate current theories and evidence for a mechanistic role of neuronal oscillations in measuring language processing, covering the latest advances brought about by EEG, MEG and fMRI imaging methods. Our main focus is to highlight innovative and foundational studies that go beyond methodological issues and advance our theoretical understanding of the role of brain oscillations in language processing. Contributions from the pioneers of this field are selected, illustrating how the study of brain oscillations has allowed investigating theoretically relevant questions that could not be addressed by more traditional methods. The topic thus aims at deepening our mechanistic understanding of language processing and bringing us closer to bridging the gap between brain, mind and behavior for the crucial cognitive function of speech. > > Hosted By: > Marcela Pena, Catholic University of Chile, Chile > Lucia Melloni, Max Planck Institute for Brain Research, Germany -------------- next part -------------- An HTML attachment was scrubbed... URL: From Santosh.VemaKrishnamurthy at nrc-cnrc.gc.ca Mon Aug 15 14:06:43 2011 From: Santosh.VemaKrishnamurthy at nrc-cnrc.gc.ca (Vema Krishnamurthy, Santosh) Date: Mon, 15 Aug 2011 08:06:43 -0400 Subject: [FieldTrip] fieldtrip Digest, Vol 9, Issue 14 In-Reply-To: References: Message-ID: Hello Gregor, Thanks for the suggestions, I will give this a try. I'm not doing forward modeling with the high resolution surfaces. I'm trying to perform surface matching technique with scalp surface and the digitization data to do the co-registration of MEG and MRI. Thanks and Regards, --------------------------------------------------- Santosh Vema KrishnaMurthy, M.A.Sc Medical Devices Engineer NRC Institute for Biodiagnostics (Atlantic) Elekta/IWK MEG Research office Ph: (902) 470-3984 --------------------------------------------------- ________________________________________ From: fieldtrip-bounces at donders.ru.nl [fieldtrip-bounces at donders.ru.nl] On Behalf Of fieldtrip-request at donders.ru.nl [fieldtrip-request at donders.ru.nl] Sent: Saturday, August 13, 2011 7:00 AM To: fieldtrip at donders.ru.nl Subject: fieldtrip Digest, Vol 9, Issue 14 Send fieldtrip mailing list submissions to fieldtrip at donders.ru.nl To subscribe or unsubscribe via the World Wide Web, visit http://mailman.science.ru.nl/mailman/listinfo/fieldtrip or, via email, send a message with subject or body 'help' to fieldtrip-request at donders.ru.nl You can reach the person managing the list at fieldtrip-owner at donders.ru.nl When replying, please edit your Subject line so it is more specific than "Re: Contents of fieldtrip digest..." Today's Topics: 1. Scalp segmentation from an MRI (.nii or .fif) (Vema Krishnamurthy, Santosh) 2. Antw: Scalp segmentation from an MRI (.nii or .fif) (Gregor Volberg) ---------------------------------------------------------------------- Message: 1 Date: Fri, 12 Aug 2011 07:00:10 -0700 From: "Vema Krishnamurthy, Santosh" To: "fieldtrip at donders.ru.nl" Subject: [FieldTrip] Scalp segmentation from an MRI (.nii or .fif) Message-ID: Content-Type: text/plain; charset="iso-8859-1" Hello Everyone, I'm very new to Fieldtrip and I need to perform scalp segmentation from an mri file(.nii or .fif). I'm working on a project where I need the scalp surface for surface matching, so I need the scalp surface with high resolution which clearly shows the facial features (nose, eyes, lips, chin). Any help/suggestions relating to this are much appreciated. Thanks and Regards, --------------------------------------------------- Santosh Vema KrishnaMurthy, M.A.Sc Medical Devices Engineer NRC Institute for Biodiagnostics (Atlantic) Elekta/IWK MEG Research office Ph: (902) 470-3984 --------------------------------------------------- -------------- next part -------------- An HTML attachment was scrubbed... URL: ------------------------------ Message: 2 Date: Fri, 12 Aug 2011 18:47:30 +0200 From: "Gregor Volberg" To: Subject: [FieldTrip] Antw: Scalp segmentation from an MRI (.nii or .fif) Message-ID: <4E457543020000570000A631 at gwsmtp1.uni-regensburg.de> Content-Type: text/plain; charset="us-ascii" Dear Santosh, for that purpose you might want to have a look at the NFT toolbox from SCCN, http://sccn.ucsd.edu/nft/ . It does high resolution segmentations for brain, csf, scalp and skin surfaces. If you need the segmentation for computing forward solutions with BEM head models, then I would discourage a too fine resolution, though. It strongly increases the computational load. Best regards, Gregor -- Dr. rer. nat. Gregor Volberg ( mailto:gregor.volberg at psychologie.uni-regensburg.de ) University of Regensburg Institute for Experimental Psychology 93040 Regensburg, Germany Tel: +49 941 943 3862 Fax: +49 941 943 3233 http://www.psychologie.uni-regensburg.de/Greenlee/team/volberg/volberg.html >>> "Vema Krishnamurthy, Santosh" 12.08.11 16.42 Uhr >>> Hello Everyone, I'm very new to Fieldtrip and I need to perform scalp segmentation from an mri file(.nii or .fif). I'm working on a project where I need the scalp surface for surface matching, so I need the scalp surface with high resolution which clearly shows the facial features (nose, eyes, lips, chin). Any help/suggestions relating to this are much appreciated. Thanks and Regards, --------------------------------------------------- Santosh Vema KrishnaMurthy, M.A.Sc Medical Devices Engineer NRC Institute for Biodiagnostics (Atlantic) Elekta/IWK MEG Research office Ph: (902) 470-3984 --------------------------------------------------- -------------- next part -------------- An HTML attachment was scrubbed... URL: ------------------------------ _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip End of fieldtrip Digest, Vol 9, Issue 14 **************************************** From shreesb at yahoo.com Mon Aug 15 21:30:22 2011 From: shreesb at yahoo.com (shree b) Date: Mon, 15 Aug 2011 12:30:22 -0700 (PDT) Subject: [FieldTrip] Header file for TDT system Message-ID: <1313436622.23112.YahooMailClassic@web162012.mail.bf1.yahoo.com> Hello, I am new to field-trip and wanted to use your analysis on 16 channel continous LFP recordings from the TDT (Tucker Davis Technology) system. As there isn't a provision to read TDT header files I tried to input the values directly into the structure array. 1)I have already imported my data into Matlab and have the information that is need to be included in cfg.headerfile in a structure array: hdr=[]; hdr.Fs=1000; hdr.nChans=16; hdr.nSamples= abfi.dataPtsPerChan; %abfi.dataPtsPerChan=620646 hdr.nSamplesPre=0; hdr.nTrials=1; hdr.label= abfi.recChNames; %abfi.recChNames is a cell array with the channel names: %IN0,IN1,IN2,IN3,IN4,IN5,IN6,IN7,IN8,IN9,IN10,IN11,IN12,IN13,IN14,IN15 2) The data is in a mat file with all 16 channels labeled as before 3) I defined trl as: a=1; b=length(IN0); c=0; trl=[a b c];% single trial no offset cfg=struct('headerfile',hdr,'datafile', data,'trl',trl) This approach isn't working, could you please tell me how I should specify the values for the header file and data structure. thanks very much Shilpa From kirihara-tky at umin.ac.jp Tue Aug 16 06:00:11 2011 From: kirihara-tky at umin.ac.jp (Kenji Kirihara) Date: Tue, 16 Aug 2011 06:00:11 JST Subject: [FieldTrip] Time-frequency analysis Message-ID: <08160600.ME2066201@umin.org> Dear FieldTrip mailing list members, I am interested in time-frequency analysis of EEG and have a question about ft_freqanalysis. I applied ft_freqanalysis to my data for time-frequency analysis using wavelets. Power at each time-requency point is calculated. However, power at each point seems to be too large. For example, power at 4 Hz, 0 ms, O2 is 1218, power at 10 Hz is 730, and power at 20 Hz is 94. These values seem to be too large compared to power calculated using FFT for frequency analysis. I would be grateful if someone tells me why this happens. Sincerely, Kenji Kirihara From BelluscB at ninds.nih.gov Mon Aug 15 23:25:21 2011 From: BelluscB at ninds.nih.gov (Belluscio, Beth (NIH/NINDS) [E]) Date: Mon, 15 Aug 2011 17:25:21 -0400 Subject: [FieldTrip] preparing the forward model in beamforming Message-ID: <794DFDD3C128EF44AC49ADC58FD0090C057456B65B@NIHMLBX10.nih.gov> Hi Fieldtrip users- I am trying to learn to use the beamforming technique as outlined in the tutorial. I have a dicom MRI that I used to create a file that contains markers for the fiducials. I then used this in ft_read_mri and ft_volumesegment - both of which seemed to work fine. Then, when I used ft_prepare_singleshell, I get this error message: "??? Undefined function or method 'imfill' for input arguments of type 'char'. Error in ==> prepare_mesh_segmentation at 68 Seg = imfill((mri.seg==cfg.tisse(i)), 'holes')" I am used Fieldtrip 20110721 Any suggestions? Beth. Beth Belluscio MD-PhD Clinical Fellow Human Motor Control Section NINDS, NIH 301-402-3495 -------------- next part -------------- An HTML attachment was scrubbed... URL: From Don.Rojas at ucdenver.edu Mon Aug 15 23:57:07 2011 From: Don.Rojas at ucdenver.edu (Rojas, Don) Date: Mon, 15 Aug 2011 15:57:07 -0600 Subject: [FieldTrip] preparing the forward model in beamforming In-Reply-To: <794DFDD3C128EF44AC49ADC58FD0090C057456B65B@NIHMLBX10.nih.gov> References: <794DFDD3C128EF44AC49ADC58FD0090C057456B65B@NIHMLBX10.nih.gov> Message-ID: <2D825D9F-FEDA-4B1C-B401-38C3878B4E36@ucdenver.edu> Beth, The matlab function imfill.m appears to be missing for your matlab install. It is part of the matlab image processing toolbox, which is not part of the base matlab package. You could check with your local matlab admin to see if you should have the image processing toolbox. FieldTrip should probably check for the existence of this toolbox when used so that it can throw a more meaningful error message. Best, Don On Aug 15, 2011, at 3:25 PM, Belluscio, Beth (NIH/NINDS) [E] wrote: Hi Fieldtrip users- I am trying to learn to use the beamforming technique as outlined in the tutorial. I have a dicom MRI that I used to create a file that contains markers for the fiducials. I then used this in ft_read_mri and ft_volumesegment – both of which seemed to work fine. Then, when I used ft_prepare_singleshell, I get this error message: “??? Undefined function or method ‘imfill’ for input arguments of type ‘char’. Error in ==> prepare_mesh_segmentation at 68 Seg = imfill((mri.seg==cfg.tisse(i)), ‘holes’)” I am used Fieldtrip 20110721 Any suggestions? Beth. Beth Belluscio MD-PhD Clinical Fellow Human Motor Control Section NINDS, NIH 301-402-3495 _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip ----------------------- Don Rojas, Ph.D. Associate Professor of Psychiatry U. of Colorado Denver Anschutz Medical Campus Director, UCD Magnetoencephalography Lab 13001 E. 17th Pl F546 Aurora, CO 80045 303-724-4994 -------------- next part -------------- An HTML attachment was scrubbed... URL: From jan.schoffelen at donders.ru.nl Tue Aug 16 10:01:28 2011 From: jan.schoffelen at donders.ru.nl (jan-mathijs schoffelen) Date: Tue, 16 Aug 2011 10:01:28 +0200 Subject: [FieldTrip] Time-frequency analysis In-Reply-To: <08160600.ME2066201@umin.org> References: <08160600.ME2066201@umin.org> Message-ID: <5FE3FAAE-6A78-4033-B84C-AEB55E844285@donders.ru.nl> Dear Kenji, In addition of applying the fft on the data, fieldtrip normalises the fft with a number that equals 2/Nsmp, where Nsmp is the length of the wavelet. Best wishes, Jan-Mathijs On Aug 15, 2011, at 11:00 PM, Kenji Kirihara wrote: > Dear FieldTrip mailing list members, > > I am interested in time-frequency analysis of EEG and have > a question about ft_freqanalysis. I applied ft_freqanalysis to > my data for time-frequency analysis using wavelets. Power at each > time-requency point is calculated. However, power at each point > seems to be too large. For example, power at 4 Hz, 0 ms, O2 is 1218, > power at 10 Hz is 730, and power at 20 Hz is 94. These values seem > to be too large compared to power calculated using FFT for frequency > analysis. I would be grateful if someone tells me why this happens. > > Sincerely, > > Kenji Kirihara > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip Dr. J.M. (Jan-Mathijs) Schoffelen Donders Institute for Brain, Cognition and Behaviour, Centre for Cognitive Neuroimaging, Radboud University Nijmegen, The Netherlands J.Schoffelen at donders.ru.nl Telephone: +31-24-3614793 From eelke.spaak at donders.ru.nl Tue Aug 16 10:36:16 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Tue, 16 Aug 2011 10:36:16 +0200 Subject: [FieldTrip] Header file for TDT system In-Reply-To: <1313436622.23112.YahooMailClassic@web162012.mail.bf1.yahoo.com> References: <1313436622.23112.YahooMailClassic@web162012.mail.bf1.yahoo.com> Message-ID: Dear Shilpa, If you already have the data and related information in Matlab, it is probably best to skip the entire trial-definition step when converting your data to FieldTrip format. It's quite straightforward to convert data into a format that FT can handle, especially if you just want it to have a single segment (or 'trial'). In your case, from the top of my head, you could initialize a structure with the following fields: data.label = abfi.recChNames; data.trial = {yourDataMatrix}; % assuming size(yourDataMatrix) == [nChans nSamples] data.fsample = 1000; data.time = {0:1/data.fsample:lastTimePoint}; data.sampleinfo = [1 nSamples]; This is just about the bare minimum that FT requires, which in your case should be sufficient. Note that both the data.trial and data.time fields should be cell arrays, with one element per trial in your data. For more detailed information about how FT raw data should look, see the following entry on our wiki: http://fieldtrip.fcdonders.nl/faq/how_are_the_various_data_structures_defined . Hope this was of help, Best, Eelke 2011/8/15 shree b : > Hello, > I am new to field-trip and wanted to use your analysis on 16 channel continous LFP recordings from the TDT (Tucker Davis Technology) system. As there isn't a provision to read TDT header files I tried to input the values directly into the structure array. > > 1)I have already imported my data into Matlab and have the information that is need to be included in cfg.headerfile in a structure array: > > hdr=[]; > > hdr.Fs=1000; > hdr.nChans=16; > hdr.nSamples= abfi.dataPtsPerChan; %abfi.dataPtsPerChan=620646 > hdr.nSamplesPre=0; > hdr.nTrials=1; > hdr.label= abfi.recChNames; > %abfi.recChNames is a cell array with the channel names:    %IN0,IN1,IN2,IN3,IN4,IN5,IN6,IN7,IN8,IN9,IN10,IN11,IN12,IN13,IN14,IN15 > > 2) The data is in a mat file with all 16 channels labeled as before > > 3) I defined trl as: > a=1; > b=length(IN0); > c=0; > > trl=[a b c];% single trial no offset > > cfg=struct('headerfile',hdr,'datafile', data,'trl',trl) > > This approach isn't working, could you please tell me how I should specify the values for the header file and data structure. > > thanks very much > Shilpa > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From jakobala at hotmail.com Tue Aug 16 14:09:05 2011 From: jakobala at hotmail.com (jakob kaiser) Date: Tue, 16 Aug 2011 14:09:05 +0200 Subject: [FieldTrip] scaling frequency data for comparison Message-ID: Dear list,I wanted to compare the outputs from different ft_freqanalysis-methods to see if they make significant differences in the final test results of my data. This were the parameters used: cfg.output = 'pow';cfg.channel = 'all';cfg.method = 'mtmconvol';cfg.taper = 'hanning';cfg.foi = 6:2:20; cfg.t_ftimwin = ones(length(cfg.foi),1).*0.5; cfg.toi = -0.5:0.05:2; WaveletConfig.method = 'wavelet'; WaveletConfig.output = 'pow'; WaveletConfig.foi = 6:2:20; WaveletConfig.toi = -0.5:0.05:2; (I chose these two methods because I found them used un papers relevant for my work)Now, as has been discussed on the list before, the scaling of the output data is very different. While the mtconvol-hanning-method gives me data points between roughly -2 an 1, with the wavelet-method I have data between roughly -800 and 600. From the (btw very helpful) tutorials and the previous list discussions I understand that this is a normal byproduct of the mathematical methods employed. However, for comparing different outputs it would be useful to scale both produced data sets to the same unit, so to speak. As both methods are supposed to measure the same thing, it should someone be possible to scale them to similar sizes, shouldn't it? Is there such a way? (Ultimately, I will have to compare the data results to results from a completely different program, which seem to have a third, different scaling altogether, so I am really puzzled how to handle the differences in scale)Generally, although this has been mentioned before, I am still not sure, if it is possible to put a certain unit to the results. Power is often reported in mV^2/Hz. If my initial data is in mV, would it be correct to say, that the hanning-method describe above delivers data in mV^2/Hz? Would this be correct for the Wavelet-method? Obviously, both cannot be true, because the scaling is extremely different. So, is it possible to specify a unit for any of these methods? Is there a way to convert the output of ft_freqanalysis-methods to this or another meaningful unit?I would be very grateful, if someone could give me a hint, how to interpret the data points here.Thank you for reading + thanks to the programmers for fieldtrip, which I like a lot Jakob -------------- next part -------------- An HTML attachment was scrubbed... URL: From eelke.spaak at donders.ru.nl Tue Aug 16 14:20:45 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Tue, 16 Aug 2011 14:20:45 +0200 Subject: [FieldTrip] scaling frequency data for comparison In-Reply-To: References: Message-ID: Dear Jakob, I can't give you any definitive answer on the power units of the different freqanalysis algorithms, but you might find the function ft_freqbaseline helpful. Absolute power values are usually not very meaningful in neuroscience research, since they depend on a lot of non-brain factors. Ft_freqbaseline performs a baseline correction (either relative, absolute, or percentage) on your power values. Across different spectral estimation methods, the relative power change should remain the same (since it is unitless). Best, Eelke 2011/8/16 jakob kaiser : > Dear list, > I wanted to compare the outputs from different ft_freqanalysis-methods to > see if they make significant differences in the final test results of my > data. This were the parameters used: > > cfg.output       = 'pow'; > cfg.channel      = 'all'; > cfg.method       = 'mtmconvol'; > cfg.taper        = 'hanning'; > cfg.foi          = 6:2:20; > cfg.t_ftimwin    = ones(length(cfg.foi),1).*0.5; > cfg.toi          = -0.5:0.05:2; > > WaveletConfig.method = 'wavelet'; > WaveletConfig.output = 'pow'; > WaveletConfig.foi = 6:2:20; > WaveletConfig.toi = -0.5:0.05:2; > (I chose these two methods because I found them used un papers relevant for > my work) > Now, as has been discussed on the list before, the scaling of the output > data is very different. While the mtconvol-hanning-method gives me data > points between roughly -2 an 1, with the wavelet-method I have data between > roughly -800 and 600. From the (btw very helpful) tutorials and the previous > list discussions I understand that this is a normal byproduct of > the mathematical methods employed. However, for comparing different outputs > it would be useful to scale both produced data sets to the same unit, so to > speak. As both methods are supposed to measure the same thing, it should > someone be possible to scale them to similar sizes, shouldn't it? Is there > such a way? (Ultimately, I will have to compare the data results to results > from a completely different program, which seem to have a third, different > scaling altogether, so I am really puzzled how to handle the differences in > scale) > Generally, although this has been mentioned before, I am still not sure, if > it is possible to put a certain unit to the results. Power is often reported > in mV^2/Hz. If my initial data is in mV, would it be correct to say, that > the hanning-method describe above delivers data in mV^2/Hz? Would this be > correct for the Wavelet-method? Obviously, both cannot be true, because the > scaling is extremely different. So, is it possible to specify a unit for any > of these methods? Is there a way to convert the output > of ft_freqanalysis-methods to this or another meaningful unit? > I would be very grateful, if someone could give me a hint, how to interpret > the data points here. > Thank you for reading + thanks to the programmers for fieldtrip, which I > like a lot > Jakob > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From kirihara-tky at umin.ac.jp Wed Aug 17 02:33:15 2011 From: kirihara-tky at umin.ac.jp (Kenji Kirihara) Date: Wed, 17 Aug 2011 02:33:15 JST Subject: [FieldTrip] Time-frequency analysis Message-ID: <08170233.ME2758201@umin.org> Dear Jan-Mathijs, Thank you for your advice. Sincerely, Kenji, In message <5FE3FAAE-6A78-4033-B84C-AEB55E844285 at donders.ru.nl> fieldtrip at donder s.ru.nl (jan-mathijs schoffelen) wrote: > From: jan-mathijs schoffelen > Date: Tue, 16 Aug 2011 10:01:28 +0200 > To: Email discussion list for the FieldTrip project > Subject: Re: [FieldTrip] Time-frequency analysis > > Dear Kenji, > > In addition of applying the fft on the data, fieldtrip normalises the fft with a number that equals 2/Nsmp, where Nsmp is the length of the wavelet. > > Best wishes, > > Jan-Mathijs > > > On Aug 15, 2011, at 11:00 PM, Kenji Kirihara wrote: > > > Dear FieldTrip mailing list members, > > > > I am interested in time-frequency analysis of EEG and have > > a question about ft_freqanalysis. I applied ft_freqanalysis to > > my data for time-frequency analysis using wavelets. Power at each > > time-requency point is calculated. However, power at each point > > seems to be too large. For example, power at 4 Hz, 0 ms, O2 is 1218, > > power at 10 Hz is 730, and power at 20 Hz is 94. These values seem > > to be too large compared to power calculated using FFT for frequency > > analysis. I would be grateful if someone tells me why this happens. > > > > Sincerely, > > > > Kenji Kirihara > > _______________________________________________ > > fieldtrip mailing list > > fieldtrip at donders.ru.nl > > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > Dr. J.M. (Jan-Mathijs) Schoffelen > Donders Institute for Brain, Cognition and Behaviour, > Centre for Cognitive Neuroimaging, > Radboud University Nijmegen, The Netherlands > J.Schoffelen at donders.ru.nl > Telephone: +31-24-3614793 > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From g.piantoni at nin.knaw.nl Wed Aug 17 12:43:23 2011 From: g.piantoni at nin.knaw.nl (Gio Piantoni) Date: Wed, 17 Aug 2011 12:43:23 +0200 Subject: [FieldTrip] Granger convention Message-ID: Hi all, The connectivity toolbox is very powerful and straightforward to use, thanks for that! However, I did not find any information about the convention used by Fieldtrip to represent asymmetrical connectivity measures (such as granger causality), f.e. in the help of the functions, in http://fieldtrip.fcdonders.nl/faq/in_what_way_can_frequency_domain_data_be_represented_in_fieldtrip nor in http://fieldtrip.fcdonders.nl/development/connectivity_tutorial Do the values in data.grangerspctrm(1, 2, :) represent the granger-causality values from channel 1 to channel 2 or the other way round? BTW, is it possible to run time-domain granger causality in Fieldtrip? Thanks! Gio -- Giovanni Piantoni, Ph.D. student Dept. Sleep & Cognition Netherlands Institute for Neuroscience Meibergdreef 47 1105 BA Amsterdam (NL) +31 (0)20 5665492 g.piantoni at nin.knaw.nl www.giovannipiantoni.com From jpnv2006 at gmail.com Wed Aug 17 14:58:07 2011 From: jpnv2006 at gmail.com (Juan Pablo Neira) Date: Wed, 17 Aug 2011 14:58:07 +0200 Subject: [FieldTrip] Forward solution using concentric spheres and BEM models Message-ID: Hello, I am working with an individual MRI and EEG data to do the forward solution with two different volume's model (concentric spheres and BEM), so at the end i can compare the results.  But i did not have the expected results of the concentric spheres model until now. This is my script: 1. Read individual MRI mri=ft_read_mri('data_patient_1\******.hdr'); %I got this "Warning: flipping 1st dimension (L-R) to obtain volume in neurological convention" %How can i do a homogenous transformation matrix, using the voxel dimensions %that are specified in hdr.dime.pixdim? 2. Realign to 'head coordinates' 3. MRI Reslicing 4. MRI segmentation: (brain, skull,scalp) 5. Create geometrical description of the brain, skull and scalp (3 compartments) cfg = []; cfg.interactive = 'no'; %segmentation method cfg.numvertices = 3000; cfg.sourceunits = 'mm'; cfg.downsample = 2; cfg.numcompartments = 3; cfg.tissue = {'brain' 'skull' 'scalp'}; cfg.smooth = 'yes'; bnd = ft_prepare_mesh(cfg, mri_segment); %The output just give me the geometrical description of the brain. How can i get the geometrical description of %the skull and scalp also to proceed to create a 3 spheres concentric sphere model? 6. If i have the 3 geometrical descriptions (brain, skull and scalp). Create a 3 concentric spheres model. cfg = []; cfg.headshape = [bnd1 bnd2 bnd3]; % brain- skull - scalp cfg.conductivity = [0.33 0.0042 0.331]; [vol, cfg] = prepare_concentricspheres(cfg); %I got the model but part of the brain is outside of its sphere and also the skull. How can I increase the radius so % the whole brain will be inside its sphere and also the skull? The sphere of the scalp fix good. I hope someone can help me solve these questions. Regards, Juan Pablo From sysoevao at psychiatry.wustl.edu Thu Aug 18 05:39:11 2011 From: sysoevao at psychiatry.wustl.edu (Sysoeva, Olga Vladimirovna) Date: Wed, 17 Aug 2011 22:39:11 -0500 Subject: [FieldTrip] mixed between/within subject Anova Message-ID: <76DC57362137854CB9592B496182D5FA0506E253FF@wusmexmbx2.medpriv.wucon.wustl.edu> Dear fieldtrip list members, I'd like to use mixed between subjects (3 groups) and within subject (2 factors each has 2 levels) Anova design to compare the ERPs using cluster permutation statistics (ft_timelockananlysis). But to my understanding it is impossible. Am I right? You comments are highly appreciated, Best Regards, Olga. Olga Sysoeva, Research Associate, PhD Washington University School of Medicine Campus Box 8134 660 South Euclid Ave Saint Louis, MO 63110-9909 The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. From sysoevao at psychiatry.wustl.edu Thu Aug 18 05:41:36 2011 From: sysoevao at psychiatry.wustl.edu (Sysoeva, Olga Vladimirovna) Date: Wed, 17 Aug 2011 22:41:36 -0500 Subject: [FieldTrip] cluster analysis for peak/latency data Message-ID: <76DC57362137854CB9592B496182D5FA0506E253FD@wusmexmbx2.medpriv.wucon.wustl.edu> Dear fieldtrip list members, I’m applying cluster permutation algorithm, implicated in Fieldtrip software to my EEG data. But I’m using it in a little bit different way. I want to cluster my data from 32 channels for peak amplitude and latency measures, not raw ERP data. So, as an input I have only chn*subjects matrix, without time dimension. Technically the script works fine (ft_timelockstatistics), and, from my point of view, it also makes perfect theoretic sense: we can examined both peak and latency of the component ( in point by point comparison it might be confusing to disentangle these effects). Certainly, the main motivation for this cluster-permutation test was to deal with multiple comparison problem and to conduct exploratory investigation (huge datasets). But in case of looking a specific component cluster permutation analysis is helping to deal with MCP introduced by multiple channels. Am I right? You comments are highly appreciated, Best Regards, Olga. Olga Sysoeva, Research Associate, PhD Washington University School of Medicine Campus Box 8134 660 South Euclid Ave Saint Louis, MO 63110-9909 The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. From e.maris at donders.ru.nl Thu Aug 18 07:40:32 2011 From: e.maris at donders.ru.nl (Eric Maris) Date: Thu, 18 Aug 2011 07:40:32 +0200 Subject: [FieldTrip] cluster analysis for peak/latency data In-Reply-To: <76DC57362137854CB9592B496182D5FA0506E253FD@wusmexmbx2.medpriv.wucon.wustl.edu> References: <76DC57362137854CB9592B496182D5FA0506E253FD@wusmexmbx2.medpriv.wucon.wustl.edu> Message-ID: <024f01cc5d69$58e406f0$0aac14d0$@maris@donders.ru.nl> Dear Olga, > But I'm using it in a little bit different way. I want to cluster my data from 32 > channels for peak amplitude and latency measures, not raw ERP data. So, as > an input I have only chn*subjects matrix, without time dimension. > Technically the script works fine (ft_timelockstatistics), and, from my point > of view, it also makes perfect theoretic sense: we can examined both peak > and latency of the component ( in point by point comparison it might be > confusing to disentangle these effects). Certainly, the main motivation for > this cluster-permutation test was to deal with multiple comparison problem > and to conduct exploratory investigation (huge datasets). But in case of > looking a specific component cluster permutation analysis is helping to deal > with MCP introduced by multiple channels. Am I right? If you do this separately for peak latency and peak amplitude, I see no problem with this approach. Best, Eric > > You comments are highly appreciated, > > Best Regards, > Olga. > > > > Olga Sysoeva, > Research Associate, PhD > Washington University School of Medicine > Campus Box 8134 > 660 South Euclid Ave > Saint Louis, MO 63110-9909 > > The materials in this message are private and may contain Protected > Healthcare Information or other information of a sensitive nature. If you are > not the intended recipient, be advised that any unauthorized use, disclosure, > copying or the taking of any action in reliance on the contents of this > information is strictly prohibited. If you have received this email in error, > please immediately notify the sender via telephone or return mail. > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From e.maris at donders.ru.nl Thu Aug 18 07:50:08 2011 From: e.maris at donders.ru.nl (Eric Maris) Date: Thu, 18 Aug 2011 07:50:08 +0200 Subject: [FieldTrip] mixed between/within subject Anova In-Reply-To: <76DC57362137854CB9592B496182D5FA0506E253FF@wusmexmbx2.medpriv.wucon.wustl.edu> References: <76DC57362137854CB9592B496182D5FA0506E253FF@wusmexmbx2.medpriv.wucon.wustl.edu> Message-ID: <025001cc5d6a$b00dc230$10294690$@maris@donders.ru.nl> Dear Olga, > I'd like to use mixed between subjects (3 groups) and within subject (2 > factors each has 2 levels) Anova design to compare the ERPs using cluster > permutation statistics (ft_timelockananlysis). > But to my understanding it is impossible. Am I right? The problem is with the interaction effects. With permutation tests, you cannot test the interaction between two (or more) between-UO variables (UO=subject or trial) nor the interaction between two or more within-UO variables. However, with permutation tests, you can test the interaction between a between- and a within-UO variables. So. In your study, you would only be unable to test the interaction between the two within-subject variables. There have been more threads on the testing of interaction effects. You can find them by searching in the discussion list archive. Best, Eric Maris > > You comments are highly appreciated, > > Best Regards, > Olga. > > > Olga Sysoeva, > Research Associate, PhD > Washington University School of Medicine > Campus Box 8134 > 660 South Euclid Ave > Saint Louis, MO 63110-9909 > > The materials in this message are private and may contain Protected > Healthcare Information or other information of a sensitive nature. If you are > not the intended recipient, be advised that any unauthorized use, disclosure, > copying or the taking of any action in reliance on the contents of this > information is strictly prohibited. If you have received this email in error, > please immediately notify the sender via telephone or return mail. > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From michael.wibral at web.de Thu Aug 18 09:57:03 2011 From: michael.wibral at web.de (Michael Wibral) Date: Thu, 18 Aug 2011 09:57:03 +0200 (CEST) Subject: [FieldTrip] mixed between/within subject Anova Message-ID: <961614915.5812596.1313654223009.JavaMail.fmail@mwmweb069> Dear Olga, you might find this reference helpful with respect to permutation tests and ANOVA-like questions: PERMUTATION TESTS FOR MULTI-FACTORIAL ANALYSIS OF VARIANCE MARTI J. ANDERSON and CAJO J. F. TER BRAAK Journal of Statistical Computation and Simulation, 2003, Vol. 73(2), pp. 85–113 Michael (let me know if you have trouble getting it) -----Ursprüngliche Nachricht----- Von: "Eric Maris" Gesendet: Aug 18, 2011 7:50:08 AM An: "'Email discussion list for the FieldTrip project'" Betreff: Re: [FieldTrip] mixed between/within subject Anova >Dear Olga, > >> I'd like to use mixed between subjects (3 groups) and within subject (2 >> factors each has 2 levels) Anova design to compare the ERPs using cluster >> permutation statistics (ft_timelockananlysis). >> But to my understanding it is impossible. Am I right? > >The problem is with the interaction effects. With permutation tests, you >cannot test the interaction between two (or more) between-UO variables >(UO=subject or trial) nor the interaction between two or more within-UO >variables. However, with permutation tests, you can test the interaction >between a between- and a within-UO variables. So. In your study, you would >only be unable to test the interaction between the two within-subject >variables. > >There have been more threads on the testing of interaction effects. You can >find them by searching in the discussion list archive. > >Best, > >Eric Maris > > > > >> >> You comments are highly appreciated, >> >> Best Regards, >> Olga. >> >> >> Olga Sysoeva, >> Research Associate, PhD >> Washington University School of Medicine >> Campus Box 8134 >> 660 South Euclid Ave >> Saint Louis, MO 63110-9909 >> >> The materials in this message are private and may contain Protected >> Healthcare Information or other information of a sensitive nature. If you >are >> not the intended recipient, be advised that any unauthorized use, >disclosure, >> copying or the taking of any action in reliance on the contents of this >> information is strictly prohibited. If you have received this email in >error, >> please immediately notify the sender via telephone or return mail. >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > >_______________________________________________ >fieldtrip mailing list >fieldtrip at donders.ru.nl >http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From Margit.Schoenherr at uk-erlangen.de Thu Aug 18 11:15:25 2011 From: Margit.Schoenherr at uk-erlangen.de (=?iso-8859-1?Q?Sch=F6nherr=2C_Margit?=) Date: Thu, 18 Aug 2011 11:15:25 +0200 Subject: [FieldTrip] Forward solution using concentric spheres and BEM models In-Reply-To: References: Message-ID: <71D8F8A56F37A947912FC86D4D9F00F4633F43AE27@XMAIL1.medads.uk-erlangen.de> Hello, I probably cannot help you with all your questions, but currently I'm also dealing with MRI segmentation and volume conductors, so maybe my solutions can be useful for you. Concerning your point 5 (geometrical description of the brain, skull, and scalp), I have done it the following way: cfg = []; cfg.method = 'segmentation'; cfg.tissue = [1 2 3]; % scalp = 1; skull = 2; brain = 3; cfg.numvertices = [2000 1000 800]; cfg.sourceunits = 'mm'; cfg.mriunits = 'mm'; bnd = ft_prepare_mesh(cfg, mri_segment); I think you have to specify a vector 'cfg.numvertices', a number of vertices for each compartment. By the way, at the beginning of ft_prepare_mesh the cfg-field 'numcompartments' is removed from the cfg structure (at least in my fieldtrip version, which is not the latest). Another question that comes to me is, why you need the segmented MRI for a concentric spheres volume conductor? So, regarding point 6, you can build such a vol-structure for example like this: vol.unit = 'm'; vol.o = origin; vol.r = [R_brain R_skull R_skin]; vol.c = [1 1/80 1]; By this, you can also define the radius as you like. Best, Margit ________________________________________ Von: fieldtrip-bounces at donders.ru.nl [fieldtrip-bounces at donders.ru.nl] im Auftrag von Juan Pablo Neira [jpnv2006 at gmail.com] Gesendet: Mittwoch, 17. August 2011 14:58 An: fieldtrip at donders.ru.nl Betreff: [FieldTrip] Forward solution using concentric spheres and BEM models Hello, I am working with an individual MRI and EEG data to do the forward solution with two different volume's model (concentric spheres and BEM), so at the end i can compare the results. But i did not have the expected results of the concentric spheres model until now. This is my script: 1. Read individual MRI mri=ft_read_mri('data_patient_1\******.hdr'); %I got this "Warning: flipping 1st dimension (L-R) to obtain volume in neurological convention" %How can i do a homogenous transformation matrix, using the voxel dimensions %that are specified in hdr.dime.pixdim? 2. Realign to 'head coordinates' 3. MRI Reslicing 4. MRI segmentation: (brain, skull,scalp) 5. Create geometrical description of the brain, skull and scalp (3 compartments) cfg = []; cfg.interactive = 'no'; %segmentation method cfg.numvertices = 3000; cfg.sourceunits = 'mm'; cfg.downsample = 2; cfg.numcompartments = 3; cfg.tissue = {'brain' 'skull' 'scalp'}; cfg.smooth = 'yes'; bnd = ft_prepare_mesh(cfg, mri_segment); %The output just give me the geometrical description of the brain. How can i get the geometrical description of %the skull and scalp also to proceed to create a 3 spheres concentric sphere model? 6. If i have the 3 geometrical descriptions (brain, skull and scalp). Create a 3 concentric spheres model. cfg = []; cfg.headshape = [bnd1 bnd2 bnd3]; % brain- skull - scalp cfg.conductivity = [0.33 0.0042 0.331]; [vol, cfg] = prepare_concentricspheres(cfg); %I got the model but part of the brain is outside of its sphere and also the skull. How can I increase the radius so % the whole brain will be inside its sphere and also the skull? The sphere of the scalp fix good. I hope someone can help me solve these questions. Regards, Juan Pablo _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From litvak.vladimir at gmail.com Thu Aug 18 11:37:59 2011 From: litvak.vladimir at gmail.com (Vladimir Litvak) Date: Thu, 18 Aug 2011 10:37:59 +0100 Subject: [FieldTrip] bdf files persisting in memory? In-Reply-To: <028001cc5d3f$45e3edb0$d1abc910$%Budd@newcastle.edu.au> References: <9231708890694273.WA.cl304kent.ac.uk@jiscmail.ac.uk> <028001cc5d3f$45e3edb0$d1abc910$%Budd@newcastle.edu.au> Message-ID: Dear Bill, This might be a memory leek in the low-level mex file. It seems that the code for this file is available so it can be fixed. I'm CCing this to the Fieldtrip list and I'll also submit a bug report to Fieldtrip developers. Best, Vladimir On Thu, Aug 18, 2011 at 1:39 AM, Bill Budd wrote: > Dear Vladimir > > I've noticed when using spm_eeg_convert to convert bdf files that the bdf > file persists in memory after the spm_eeg_convert function or even the > script calling it has finished. Not sure if the issue is specific to my > scripting or the spm/fieldtrip functions. Wondered if you have any advice as > it uses a lot of memory while the rest of the script pipeline runs. The only > way I can seem to clear it is to run 'clear all mex' or exit matlab, but > really need some less sledge-hammer style solution to clearing it from RAM. > > Cheers > -Bill > > From drivolta81 at gmail.com Thu Aug 18 15:18:19 2011 From: drivolta81 at gmail.com (Davide Rivolta) Date: Thu, 18 Aug 2011 15:18:19 +0200 Subject: [FieldTrip] common filters - beamforming Message-ID: Dear Fieldtrippers, I am trying to learn beamforming using the tutorial. I am succesful in calculating, with my own data, the relative change between the stuimulus and the baseline condition (as shown in the tutorial). However, if I want to do statistic (ie., compare the baseline and stimulus condition) I would need a common filter (not in the tutorial). I found very useful the website where this step is explained (examples of matlab scripts section). I am however confused on how to perform stats at the source level. I get the 2 conditions separately and I wish to compare them and hopefully plot WHERE the difference is seen. Do you have any advice on how to do it (or where to look)? Here is my script and I am sure I am doing some silly mistake in the last (statistic part) since it gives the error: ??? Reference to non-existent field 'pow'. cfg = []; cfg.toilim = [-0.5 0]; dataPre = ft_redefinetrial(cfg, DataOut); cfg.toilim = [0.5 1]; dataPost = ft_redefinetrial(cfg, DataOut); data = ft_appenddata([], dataPre, dataPost); design = [ones(1, length(dataPre.trial)) ones(1, length(dataPost.trial))*2]; cfg = []; cfg.method = 'mtmfft'; cfg.output = 'powandcsd'; cfg.taper = 'dpss'; cfg.keeptrials = 'yes'; cfg.keeptapers = 'no'; cfg.foi = 57:2:63; cfg.tapsmofrq = ones(length(cfg.foi), 1).* 5; freq = ft_freqanalysis(cfg, data); cfg = []; cfg.frequency = 60; cfg.method = 'dics'; cfg.grid = grid; cfg.vol = vol; cfg.grad = DataOut.hdr.grad; cfg.keepfilter = 'yes'; source = ft_sourceanalysis(cfg, freq); cfg = []; cfg.frequency = 60; cfg.method = 'dics'; cfg.grid = grid; cfg.vol = vol; cfg.grad = DataOut.hdr.grad; cfg.grid.filter = source.avg.filter; %uses a precomputed filter (common filter) cfg.rawtrial = 'yes'; source = ft_sourceanalysis(cfg, freq); A = find(design==1); B = find(design==2); sourceA = source; sourceA.trial(B) = []; sourceA.cumtapcnt(B) = []; sourceA.df = length(A); cfg = []; cfg.keeptrials = 'yes'; sourceA = ft_sourcedescriptives([], sourceA); sourceB = source; sourceB.trial(A) = []; sourceB.cumtapcnt(A) = []; sourceB.df = length(B); cfg = []; cfg.keeptrials = 'yes'; sourceB = ft_sourcedescriptives([], sourceB); %STATS cfg = []; cfg.parameter = 'pow'; cfg.method = 'analytic'; cfg.statistic = 'actvsblT'; cfg.computestat = 'yes'; cfg.computecritval = 'yes'; cfg.computeprob = 'yes'; cfg.alpha = 0.05; cfg.tail = 0; stat = ft_sourcestatistics(cfg, sourceB, sourceA); Thanks for your help, Davide -------------- next part -------------- An HTML attachment was scrubbed... URL: From sysoevao at psychiatry.wustl.edu Thu Aug 18 18:02:11 2011 From: sysoevao at psychiatry.wustl.edu (Sysoeva, Olga Vladimirovna) Date: Thu, 18 Aug 2011 11:02:11 -0500 Subject: [FieldTrip] mixed between/within subject Anova In-Reply-To: <961614915.5812596.1313654223009.JavaMail.fmail@mwmweb069> References: <961614915.5812596.1313654223009.JavaMail.fmail@mwmweb069> Message-ID: <76DC57362137854CB9592B496182D5FA0506F6EAFB@wusmexmbx2.medpriv.wucon.wustl.edu> Thank you Michael, Thank you Eric Maris, So, as far as I understand there are some suggestion (Anderson et al, 2003) about how to apply the permutation test to my case (1 between and 2 within subjects factors), but it is not implemented in fieldtrip yet. As for examination the interaction for 1 between and 1 within subject factors - it is possible to do with fieldtrip scripts, isn't it? I've looked through the archive and one of the option seemed to be just to examine the between subject effect on the data obtained from subtraction of one level of within subject variable from the another one ( in case of 2 levels). And the results will represent the interaction effect. Am I right? Best Regards, Olga. -----Original Message----- From: fieldtrip-bounces at donders.ru.nl [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Michael Wibral Sent: Thursday, August 18, 2011 2:57 AM To: Email discussion list for the FieldTrip project Subject: Re: [FieldTrip] mixed between/within subject Anova Dear Olga, you might find this reference helpful with respect to permutation tests and ANOVA-like questions: PERMUTATION TESTS FOR MULTI-FACTORIAL ANALYSIS OF VARIANCE MARTI J. ANDERSON and CAJO J. F. TER BRAAK Journal of Statistical Computation and Simulation, 2003, Vol. 73(2), pp. 85–113 Michael (let me know if you have trouble getting it) -----Ursprüngliche Nachricht----- Von: "Eric Maris" Gesendet: Aug 18, 2011 7:50:08 AM An: "'Email discussion list for the FieldTrip project'" Betreff: Re: [FieldTrip] mixed between/within subject Anova >Dear Olga, > >> I'd like to use mixed between subjects (3 groups) and within subject >> (2 factors each has 2 levels) Anova design to compare the ERPs using >> cluster permutation statistics (ft_timelockananlysis). >> But to my understanding it is impossible. Am I right? > >The problem is with the interaction effects. With permutation tests, >you cannot test the interaction between two (or more) between-UO >variables (UO=subject or trial) nor the interaction between two or more >within-UO variables. However, with permutation tests, you can test the >interaction between a between- and a within-UO variables. So. In your >study, you would only be unable to test the interaction between the two >within-subject variables. > >There have been more threads on the testing of interaction effects. You >can find them by searching in the discussion list archive. > >Best, > >Eric Maris > > > > >> >> You comments are highly appreciated, >> >> Best Regards, >> Olga. >> >> >> Olga Sysoeva, >> Research Associate, PhD >> Washington University School of Medicine Campus Box 8134 >> 660 South Euclid Ave >> Saint Louis, MO 63110-9909 >> >> The materials in this message are private and may contain Protected >> Healthcare Information or other information of a sensitive nature. If >> you >are >> not the intended recipient, be advised that any unauthorized use, >disclosure, >> copying or the taking of any action in reliance on the contents of >> this information is strictly prohibited. If you have received this >> email in >error, >> please immediately notify the sender via telephone or return mail. >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > >_______________________________________________ >fieldtrip mailing list >fieldtrip at donders.ru.nl >http://mailman.science.ru.nl/mailman/listinfo/fieldtrip _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. From sust2001 at yahoo.ca Thu Aug 18 21:22:56 2011 From: sust2001 at yahoo.ca (Mark Taylor) Date: Thu, 18 Aug 2011 12:22:56 -0700 (PDT) Subject: [FieldTrip] Need your help to run *.ntt data file in the FieldTrip Message-ID: <1313695376.13817.YahooMailNeo@web46410.mail.sp1.yahoo.com> Hi, Could you please let me know whether I can  run the spike data file recorded with *.ntt format from Neuralynx in the FieldTrip environment? As this *.ntt format is not listed in the option of supported spike and LFP data formats, I would be very grateful if I can get your valuable suggestions to make the FieldTrip useful for my recorded data analysis. Your kind help and the earliest reply would be really appreciated. Thanking you, Mark. -------------- next part -------------- An HTML attachment was scrubbed... URL: From Ulrich.Pomper at charite.de Fri Aug 19 12:27:20 2011 From: Ulrich.Pomper at charite.de (Pomper, Ulrich) Date: Fri, 19 Aug 2011 12:27:20 +0200 Subject: [FieldTrip] =?windows-1252?q?1_PhD_and_1_Postdoctoral_Position_-_?= =?windows-1252?q?Charit=E9_Berlin?= Message-ID: The Department of Psychiatry and Psychotherapy, Charité – Universitätsmedizin Berlin (http://psy-ccm.charite.de/en/) invites applications for a Post-doctoral and a PhD student position. A Starting Grant of the European Research Council (ERC) will fund both positions for initially 2 years with the possibility of extension. The main objective of this ERC research program is to examine neural and neurochemical markers of multisensory integration and to test a new hypothesis that considers dynamic interplay of synchronized neural populations as a key to multisensory processes (Senkowski et al. 2008, Trends in Neurosciences). The studies within this program include healthy subjects and patients with schizophrenia, as a prototype of a mental disorder with deficits in multisensory integration. Multisensory processes will be examined in a series of experiments requiring both bottom-up and top-down processing (Talsma et al. 2010, Trends in Cognitive Sciences). Reaching beyond the state-of-the-art, this comprises a combination of human EEG/MEG data as a macroscopic measure of cortical processing; source modeling of synchronized EEG/MEG activity; and neurochemical markers using MR spectroscopy. Applicants should have a background in psychology, medicine, biology, physics or neuroscience. Experience in human EEG/MEG studies or biosignal analysis (e.g., Matlab) is desirable (i.e. required for the Post-doctoral position). An interest in neurophysiologic studies in clinical populations is expected, as well as basic German language skills for interacting with patients. Applicants are asked to submit their CV, a short motivation letter, 2 names of referees, and documentation of relevant qualification (e.g., copies of diplomas and/or transcripts of grades) until September 15, 2011, electronically to Dr. Daniel Senkowski (daniel.senkowski at charite.de; www.danielsenkowski.com). --------------------------------------------------------------- Daniel Senkowski, Ph.D Head of the research group Multisensory-Mind Department of Psychiatry and Psychotherapy Charité, University Medicine Berlin St. Hedwig Hospital, Große Hamburger Str. 5-11 10115 Berlin, Germany Phone: +49-30-2311-2738 Fax: +49-30-2311-2750 www.danielsenkowski.com From jpnv2006 at gmail.com Fri Aug 19 16:01:16 2011 From: jpnv2006 at gmail.com (Juan Pablo Neira) Date: Fri, 19 Aug 2011 16:01:16 +0200 Subject: [FieldTrip] Forward solution using concentric spheres and BEM models In-Reply-To: <71D8F8A56F37A947912FC86D4D9F00F4633F43AE27@XMAIL1.medads.uk-erlangen.de> References: <71D8F8A56F37A947912FC86D4D9F00F4633F43AE27@XMAIL1.medads.uk-erlangen.de> Message-ID: Thank you Margit for your help, I tried your sugestion to create the geometrical description of the brain, skull and scalp but I am still having the same output bnd = 1x1 struct (pnt and tri just of the brain). I should have bnd = 1x3 struct (pnt and tri of the brain, skull and scalp). This is the information in the command line in matlab not downsampling brain not downsampling scalp not downsampling skull using the segmentation approach using the segmented MRI triangulating the boundary of compartment 1 segmentation compartment 1 of 1 completed Can you tell me which version of fieldtrip (fieldtrip-yyyymmdd) are you using so i could compare the code of the function ft_prepare_mesh. Regards, Juan Pablo 2011/8/18 Schönherr, Margit : > Hello, > > I probably cannot help you with all your questions, but currently I'm also dealing with MRI segmentation and volume conductors, so maybe my solutions can be useful for you. > Concerning your point 5 (geometrical description of the brain, skull, and scalp), I have done it the following way: > > cfg             = []; > cfg.method      = 'segmentation'; > cfg.tissue      = [1 2 3]; % scalp = 1; skull = 2; brain = 3; > cfg.numvertices = [2000 1000 800]; > cfg.sourceunits = 'mm'; > cfg.mriunits    = 'mm'; > bnd = ft_prepare_mesh(cfg, mri_segment); > > I think you have to specify a vector 'cfg.numvertices', a number of vertices for each compartment. > By the way, at the beginning of ft_prepare_mesh the cfg-field 'numcompartments' is removed from the cfg structure (at least in my fieldtrip version, which is not the latest). > > Another question that comes to me is, why you need the segmented MRI for a concentric spheres volume conductor? So, regarding point 6, you can build such a vol-structure for example like this: > > vol.unit = 'm'; > vol.o = origin; > vol.r = [R_brain R_skull R_skin]; > vol.c = [1 1/80 1]; > > By this, you can also define the radius as you like. > > Best, > Margit > > > > > ________________________________________ > Von: fieldtrip-bounces at donders.ru.nl [fieldtrip-bounces at donders.ru.nl] im Auftrag von Juan Pablo Neira [jpnv2006 at gmail.com] > Gesendet: Mittwoch, 17. August 2011 14:58 > An: fieldtrip at donders.ru.nl > Betreff: [FieldTrip] Forward solution using concentric spheres and BEM models > > Hello, > > I am working with an individual MRI and EEG data to do the forward > solution with two different volume's model (concentric spheres and > BEM), > so at the end i can compare the results.  But i did not have the > expected results of the concentric spheres model until now. > > This is my script: > > 1.  Read individual MRI > mri=ft_read_mri('data_patient_1\******.hdr'); > > %I got this "Warning: flipping 1st dimension (L-R) to obtain volume in > neurological convention" > > %How can i do a homogenous transformation matrix, using the voxel dimensions > %that are specified in hdr.dime.pixdim? > > 2. Realign to 'head coordinates' > > 3. MRI Reslicing > > 4. MRI segmentation: (brain, skull,scalp) > > 5. Create geometrical description of the brain, skull and scalp (3 compartments) > > cfg                 = []; > cfg.interactive     = 'no';     %segmentation method > cfg.numvertices     = 3000; > cfg.sourceunits     = 'mm'; > cfg.downsample      = 2; > cfg.numcompartments = 3; > cfg.tissue          = {'brain' 'skull' 'scalp'}; > cfg.smooth          = 'yes'; > bnd                 = ft_prepare_mesh(cfg, mri_segment); > > %The output just give me the geometrical description of the brain. > How can i get the geometrical description of > %the skull and scalp also to proceed to create a 3 spheres concentric > sphere model? > > 6.  If i have the 3 geometrical descriptions (brain, skull and scalp). >  Create a 3 concentric spheres model. > > cfg                 = []; > cfg.headshape       = [bnd1 bnd2 bnd3];   % brain- skull - scalp > cfg.conductivity    = [0.33 0.0042 0.331]; > [vol, cfg]          = prepare_concentricspheres(cfg); > > %I got the model but part of the brain is outside of its sphere and > also the skull.  How can I increase the radius so > % the whole brain will be inside its sphere and also the skull?  The > sphere of the scalp fix good. > > I hope someone can help me solve these questions. > > Regards, > > Juan Pablo > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From Margit.Schoenherr at uk-erlangen.de Fri Aug 19 17:23:34 2011 From: Margit.Schoenherr at uk-erlangen.de (=?iso-8859-1?Q?Sch=F6nherr=2C_Margit?=) Date: Fri, 19 Aug 2011 17:23:34 +0200 Subject: [FieldTrip] Forward solution using concentric spheres and BEM models In-Reply-To: References: <71D8F8A56F37A947912FC86D4D9F00F4633F43AE27@XMAIL1.medads.uk-erlangen.de>, Message-ID: <71D8F8A56F37A947912FC86D4D9F00F4633F43AE2A@XMAIL1.medads.uk-erlangen.de> Hello Juan Pablo, I am using fieldtrip-20110603. Now that you say, that your problem persisted, I remember having changed the code of prepare_mesh_segmentation. You find the function in the private directory. In line 25, it sets cfg.tissue = 1 if mri has fields 'brain' or 'scalp'. I think this explains why the segmentation is done for compartment 1 only. So I have simply commented this line, so that cfg.tissue is left unchanged, and the segmentation is done for all compartments. I know that this is bad style, but I wanted to get it working. And since I was relatively new to fieldtrip, I hesitated to ask the discussion list... I hope this helps. And I'm open for better solutions. Best, Margit ________________________________________ Von: fieldtrip-bounces at donders.ru.nl [fieldtrip-bounces at donders.ru.nl] im Auftrag von Juan Pablo Neira [jpnv2006 at gmail.com] Gesendet: Freitag, 19. August 2011 16:01 An: Email discussion list for the FieldTrip project Betreff: Re: [FieldTrip] Forward solution using concentric spheres and BEM models Thank you Margit for your help, I tried your sugestion to create the geometrical description of the brain, skull and scalp but I am still having the same output bnd = 1x1 struct (pnt and tri just of the brain). I should have bnd = 1x3 struct (pnt and tri of the brain, skull and scalp). This is the information in the command line in matlab not downsampling brain not downsampling scalp not downsampling skull using the segmentation approach using the segmented MRI triangulating the boundary of compartment 1 segmentation compartment 1 of 1 completed Can you tell me which version of fieldtrip (fieldtrip-yyyymmdd) are you using so i could compare the code of the function ft_prepare_mesh. Regards, Juan Pablo 2011/8/18 Schönherr, Margit : > Hello, > > I probably cannot help you with all your questions, but currently I'm also dealing with MRI segmentation and volume conductors, so maybe my solutions can be useful for you. > Concerning your point 5 (geometrical description of the brain, skull, and scalp), I have done it the following way: > > cfg = []; > cfg.method = 'segmentation'; > cfg.tissue = [1 2 3]; % scalp = 1; skull = 2; brain = 3; > cfg.numvertices = [2000 1000 800]; > cfg.sourceunits = 'mm'; > cfg.mriunits = 'mm'; > bnd = ft_prepare_mesh(cfg, mri_segment); > > I think you have to specify a vector 'cfg.numvertices', a number of vertices for each compartment. > By the way, at the beginning of ft_prepare_mesh the cfg-field 'numcompartments' is removed from the cfg structure (at least in my fieldtrip version, which is not the latest). > > Another question that comes to me is, why you need the segmented MRI for a concentric spheres volume conductor? So, regarding point 6, you can build such a vol-structure for example like this: > > vol.unit = 'm'; > vol.o = origin; > vol.r = [R_brain R_skull R_skin]; > vol.c = [1 1/80 1]; > > By this, you can also define the radius as you like. > > Best, > Margit > > > > > ________________________________________ > Von: fieldtrip-bounces at donders.ru.nl [fieldtrip-bounces at donders.ru.nl] im Auftrag von Juan Pablo Neira [jpnv2006 at gmail.com] > Gesendet: Mittwoch, 17. August 2011 14:58 > An: fieldtrip at donders.ru.nl > Betreff: [FieldTrip] Forward solution using concentric spheres and BEM models > > Hello, > > I am working with an individual MRI and EEG data to do the forward > solution with two different volume's model (concentric spheres and > BEM), > so at the end i can compare the results. But i did not have the > expected results of the concentric spheres model until now. > > This is my script: > > 1. Read individual MRI > mri=ft_read_mri('data_patient_1\******.hdr'); > > %I got this "Warning: flipping 1st dimension (L-R) to obtain volume in > neurological convention" > > %How can i do a homogenous transformation matrix, using the voxel dimensions > %that are specified in hdr.dime.pixdim? > > 2. Realign to 'head coordinates' > > 3. MRI Reslicing > > 4. MRI segmentation: (brain, skull,scalp) > > 5. Create geometrical description of the brain, skull and scalp (3 compartments) > > cfg = []; > cfg.interactive = 'no'; %segmentation method > cfg.numvertices = 3000; > cfg.sourceunits = 'mm'; > cfg.downsample = 2; > cfg.numcompartments = 3; > cfg.tissue = {'brain' 'skull' 'scalp'}; > cfg.smooth = 'yes'; > bnd = ft_prepare_mesh(cfg, mri_segment); > > %The output just give me the geometrical description of the brain. > How can i get the geometrical description of > %the skull and scalp also to proceed to create a 3 spheres concentric > sphere model? > > 6. If i have the 3 geometrical descriptions (brain, skull and scalp). > Create a 3 concentric spheres model. > > cfg = []; > cfg.headshape = [bnd1 bnd2 bnd3]; % brain- skull - scalp > cfg.conductivity = [0.33 0.0042 0.331]; > [vol, cfg] = prepare_concentricspheres(cfg); > > %I got the model but part of the brain is outside of its sphere and > also the skull. How can I increase the radius so > % the whole brain will be inside its sphere and also the skull? The > sphere of the scalp fix good. > > I hope someone can help me solve these questions. > > Regards, > > Juan Pablo > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From P.Praamstra at neuro.umcn.nl Fri Aug 19 21:04:29 2011 From: P.Praamstra at neuro.umcn.nl (P.Praamstra at neuro.umcn.nl) Date: Fri, 19 Aug 2011 21:04:29 +0200 Subject: [FieldTrip] PhD and post-doctoral position at Donders Centre for Cognitive Neuroimaging Message-ID: Please find attached two adverts for positions at the Donders Centre for Cognitive Neuroimaging, Radboud University Nijmegen, The Netherlands. For both positions we are looking for candidates with MEG or EEG experience. Post-doctoral researcher on project "Movement selection in the parietofrontal cortex: The affordance competition hypothesis". PhD student for project "Feeling the beat: The neurophysiology of cueing in Parkinson's disease". Peter Praamstra, MD, PhD Department of Neurology Radboud University Nijmegen Medical Centre PO Box 9101, 6500 HB Nijmegen The Netherlands Tel: 00-31-24-3668254 Fax: 00-31-24-3541122 Email: p.praamstra at neuro.umcn.nl Het UMC St Radboud staat geregistreerd bij de Kamer van Koophandel in het handelsregister onder nummer 41055629. The Radboud University Nijmegen Medical Centre is listed in the Commercial Register of the Chamber of Commerce under file number 41055629. -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: Advert_Post-doc.doc Type: application/msword Size: 33280 bytes Desc: Advert_Post-doc.doc URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: Advert_AiO.doc Type: application/msword Size: 33280 bytes Desc: Advert_AiO.doc URL: From lmoranr at gmail.com Tue Aug 23 12:02:33 2011 From: lmoranr at gmail.com (=?ISO-8859-1?Q?Luis_Mor=E1n?=) Date: Tue, 23 Aug 2011 12:02:33 +0200 Subject: [FieldTrip] Doubt about de units Message-ID: Dear fieldtrippers, I'm doing a simulation with somato data from MEG. I know that measured fields in data are in* femtoteslas*. Well, the procedure is: - Load data - After setting the configuration structure "cfg", I make a dipolesimulation, and then timelockanalysis, i.e.: *raw1 = dipolesimulation(cfg) avg1=timelockanalysis(cfg, raw1)* What I wonder to know is which are the resultant units of the field * avg1.avg* In the same way, then I perform a reconstruction using beamforming: *sourceAvg1 = sourceanalysis(cfg_beamforming,avg1);* And also, I would like to know the units in which the field * sourceAvg1.avg.pow* is measured Thanks in advance Regards -- Luis Morán Rodríguez -------------- next part -------------- An HTML attachment was scrubbed... URL: From Antony.Passaro at uth.tmc.edu Tue Aug 23 20:50:47 2011 From: Antony.Passaro at uth.tmc.edu (Passaro, Antony D) Date: Tue, 23 Aug 2011 13:50:47 -0500 Subject: [FieldTrip] Source Timecourse In-Reply-To: References: Message-ID: Hi Fieldtrippers, I was looking for a way to extract a time-course from each source estimated by the lcmv beamformer. I searched through the mailing list archives and came across a few emails which talked about the possibility but I was unable to find any examples explaining exactly how this process is done. I am interested in extracting the time-course over the average response rather than the individual trials, if possible. Any help would be greatly appreciated. Thanks, -Tony -------------- next part -------------- An HTML attachment was scrubbed... URL: From Don.Rojas at ucdenver.edu Thu Aug 25 19:43:03 2011 From: Don.Rojas at ucdenver.edu (Rojas, Don) Date: Thu, 25 Aug 2011 11:43:03 -0600 Subject: [FieldTrip] Source Timecourse In-Reply-To: References: Message-ID: <51AF637A-B1EB-4421-84B3-6B0D0C4BBA24@ucdenver.edu> Tony, I don't use the LCMV beamformer, so I can't tell you what the structure of the output is, but what you need to perform a source space projection is the orientation and location information for the sources. In the DICS beamformer output, which I'm familiar with, this information would be in the source structure as follows (assuming your source output structure is called "source"): source.pos % the position info as an n x 3 matrix, where n is the number of sources, in the same units as your volume conductor model source.avg.ori % the orientation as an 1 x n cell array, where n is the number of locations and each cell is a 3 x 1 array of x y z orientation and n refers to the % same source in each array So, in that case, you might find the necessary information for the peak source, in my case, as follows: [~,ind]=max(source.avg.pow(:)); % index of the source with the maximum power in the dics output, let's say ind = 5000 ori=source.avg.ori(5000); % orientation info loc=source.pos(5000,:); % location information Then, if you get this info, you can search the archives for my posts, one of which has a detailed explanation for getting the source waveform when you have a known source location, source orientation and a volume conductor. If you want to get all the waveforms, you can of course iterate the process over the entire voxel volume, but since the leadfields are highly correlated in adjacent source locations, you will see that many of the waveforms will look essentially the same for sources within a few centimeters. Best, Don ----------------------- Don Rojas, Ph.D. Associate Professor of Psychiatry U. of Colorado Denver Anschutz Medical Campus Director, UCD Magnetoencephalography Lab 13001 E. 17th Pl F546 Aurora, CO 80045 USA 303-724-4994 On Aug 23, 2011, at 12:50 PM, Passaro, Antony D wrote: Hi Fieldtrippers, I was looking for a way to extract a time-course from each source estimated by the lcmv beamformer. I searched through the mailing list archives and came across a few emails which talked about the possibility but I was unable to find any examples explaining exactly how this process is done. I am interested in extracting the time-course over the average response rather than the individual trials, if possible. Any help would be greatly appreciated. Thanks, -Tony _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip -------------- next part -------------- An HTML attachment was scrubbed... URL: From ekanal at cmu.edu Fri Aug 26 18:19:45 2011 From: ekanal at cmu.edu (Kanal Eliezer) Date: Fri, 26 Aug 2011 12:19:45 -0400 Subject: [FieldTrip] assessing significance in using ft_timelockanalysis results Message-ID: <8FC0FE65-8E0D-4D7C-9A62-CE1610845C92@cmu.edu> Hello folks - The event related statistics tutorial (http://fieldtrip.fcdonders.nl/tutorial/eventrelatedstatistics) talks about assessing significance parametrically by running t-tests on pooled timelockanalysis data. My question is, does the fact that the averages were created from N trials make a difference? If I'm condition A has twelve averages and condition B has another twelve, and each average contains 70 trials, is there a way to "inform" the statistical test that the power in this dataset is greater than 24? Is this only possible if I run the t-test comparing each set of 840 (70*12) trials? I'm also curious whether this is possible with non-parametric analyses, as well. Thanks - Elli Kanal -------------------- Eliezer Kanal, Ph.D. Postdoctoral Fellow Center for the Neural Basis of Cognition Carnegie Mellon University 4400 Fifth Ave, Suite 110A Pittsburgh PA 15213 P: 412-268-4115 F: 412-268-5060 From e.maris at donders.ru.nl Fri Aug 26 21:00:18 2011 From: e.maris at donders.ru.nl (Eric Maris) Date: Fri, 26 Aug 2011 21:00:18 +0200 Subject: [FieldTrip] assessing significance in using ft_timelockanalysis results In-Reply-To: <8FC0FE65-8E0D-4D7C-9A62-CE1610845C92@cmu.edu> References: <8FC0FE65-8E0D-4D7C-9A62-CE1610845C92@cmu.edu> Message-ID: <020701cc6422$65b10f30$31132d90$@maris@donders.ru.nl> Hi Kanal, > The event related statistics tutorial > (http://fieldtrip.fcdonders.nl/tutorial/eventrelatedstatistics) talks about > assessing significance parametrically by running t-tests on pooled > timelockanalysis data. My question is, does the fact that the averages were > created from N trials make a difference? If I'm condition A has twelve > averages and condition B has another twelve, and each average contains 70 > trials, is there a way to "inform" the statistical test that the power in this > dataset is greater than 24? Is this only possible if I run the t-test comparing > each set of 840 (70*12) trials? > > I'm also curious whether this is possible with non-parametric analyses, as > well. Thanks - In an analysis over subjects (called random-effects analysis in the fMRI literature), "informing" the statistical test about the number of trials per condition only makes sense if this number is different for the two conditions. I propose that you have a look the fMRI papers that deal with the issue of fixed-versus-random effect analyses. The conceptual issues involved are the same in fMRI and electrophysiology. Best, Eric Maris > > Elli Kanal > > > -------------------- > Eliezer Kanal, Ph.D. > Postdoctoral Fellow > Center for the Neural Basis of Cognition > Carnegie Mellon University > 4400 Fifth Ave, Suite 110A > Pittsburgh PA 15213 > P: 412-268-4115 > F: 412-268-5060 > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From tim.curran at Colorado.EDU Sat Aug 27 21:31:25 2011 From: tim.curran at Colorado.EDU (Tim Curran) Date: Sat, 27 Aug 2011 13:31:25 -0600 Subject: [FieldTrip] Fwd: Postdoctoral Position Job Announcement for Bellugi at Salk Institute References: <34FBC147-2620-42D5-ACE8-0D5FD88AC5BD@eng.ucsd.edu> Message-ID: <1686575C-9B5C-4ACB-B752-EF360CF1B8B0@colorado.edu> > > > > -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: Job Ad Final Version_06_14_11.pdf Type: application/pdf Size: 728679 bytes Desc: Job Ad Final Version_06_14_11.pdf URL: From r.oostenveld at donders.ru.nl Mon Aug 29 13:24:21 2011 From: r.oostenveld at donders.ru.nl (Robert Oostenveld) Date: Mon, 29 Aug 2011 13:24:21 +0200 Subject: [FieldTrip] Source Timecourse In-Reply-To: <51AF637A-B1EB-4421-84B3-6B0D0C4BBA24@ucdenver.edu> References: <51AF637A-B1EB-4421-84B3-6B0D0C4BBA24@ucdenver.edu> Message-ID: Dear Tony The lcmv beamformer (by default) estimates the time series of the source, given that the source has an unknown orientation. The result is a 3xNtime matrix as the source activation, which is in source.avg.mom (or in source.trial(i).mom). The three rows of the moment are the strength of the source in the x-, y- and z-direction. Using ft_sourcedescriptives with the option cfg.projectmom=yes you can project the source to its strongest orientation, i.e. the direction that explains most of the source variance. That is equivalent to taking the largest eigenvector of the source timeseries (which is a phrasing that is often used in papers, also on combining fMRI bold timeseries over multiple voxels). You can also do it manually, like this mom3d = randn(3,100); % example source dipole moment [u, s, v] = svd(mom, 'econ'); mom1d = v(1,:); Alternative to taking the largest eigenvector (which has a sign-ambiguity), you can do something like mom3d = randn(3,100); % example source dipole moment [u, s, v] = svd(mom, 'econ'); mom1d = abs(v(1,:)); to take the absolute value, or mom3d = randn(3,100); % example source dipole moment mom1d = sqrt(sum(mom3d.^2,1)); to take the strength over all three directions for each timepoint. best Robert On 25 Aug 2011, at 19:43, Rojas, Don wrote: > Tony, > > I don't use the LCMV beamformer, so I can't tell you what the structure of the output is, but what you need to perform a source space projection is the orientation and location information for the sources. In the DICS beamformer output, which I'm familiar with, this information would be in the source structure as follows (assuming your source output structure is called "source"): > > source.pos % the position info as an n x 3 matrix, where n is the number of sources, in the same units as your volume conductor model > source.avg.ori % the orientation as an 1 x n cell array, where n is the number of locations and each cell is a 3 x 1 array of x y z orientation and n refers to the % same source in each array > > So, in that case, you might find the necessary information for the peak source, in my case, as follows: > > [~,ind]=max(source.avg.pow(:)); % index of the source with the maximum power in the dics output, let's say ind = 5000 > ori=source.avg.ori(5000); % orientation info > loc=source.pos(5000,:); % location information > > Then, if you get this info, you can search the archives for my posts, one of which has a detailed explanation for getting the source waveform when you have a known source location, source orientation and a volume conductor. If you want to get all the waveforms, you can of course iterate the process over the entire voxel volume, but since the leadfields are highly correlated in adjacent source locations, you will see that many of the waveforms will look essentially the same for sources within a few centimeters. > > Best, > > Don > ----------------------- > Don Rojas, Ph.D. > Associate Professor of Psychiatry > U. of Colorado Denver Anschutz Medical Campus > Director, UCD Magnetoencephalography Lab > 13001 E. 17th Pl F546 > Aurora, CO 80045 USA > 303-724-4994 > > On Aug 23, 2011, at 12:50 PM, Passaro, Antony D wrote: > >> Hi Fieldtrippers, >> >> I was looking for a way to extract a time-course from each source estimated by the lcmv beamformer. I searched through the mailing list archives and came across a few emails which talked about the possibility but I was unable to find any examples explaining exactly how this process is done. I am interested in extracting the time-course over the average response rather than the individual trials, if possible. Any help would be greatly appreciated. >> >> Thanks, >> -Tony >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From Antony.Passaro at uth.tmc.edu Mon Aug 29 18:33:55 2011 From: Antony.Passaro at uth.tmc.edu (Passaro, Antony D) Date: Mon, 29 Aug 2011 11:33:55 -0500 Subject: [FieldTrip] Source Timecourse In-Reply-To: References: <51AF637A-B1EB-4421-84B3-6B0D0C4BBA24@ucdenver.edu> Message-ID: Hi Robert and Don, Thank you both for your feedback, I really appreciate it. Using the ft_sourcedescriptives function (and projectmom), I was able to extract the time-course for each source. I am still curious about the output though as it appears to represent the entire duration (including the baseline) of the epoch rather than the specified time-window defined during timelockanalysis (just prior to sourceanalysis). That being the case, does this time-course represent the .pow estimated by the lcmv beamformer or is the nai? Or is it simply a source representation of the gradiometers over the entire epoch? Another issue I came across concerns the source interpolation. It seems once I interpolate the sources using a template mri, only the pow and nai fields are interpolated....is there a way to also interpolate the mom field as well or is it a matter of backtracking to figure out which original source corresponds to each interpolated source? Thank you again for your help, -Tony -----Original Message----- From: fieldtrip-bounces at donders.ru.nl [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Robert Oostenveld Sent: Monday, August 29, 2011 6:24 AM To: Email discussion list for the FieldTrip project Subject: Re: [FieldTrip] Source Timecourse Dear Tony The lcmv beamformer (by default) estimates the time series of the source, given that the source has an unknown orientation. The result is a 3xNtime matrix as the source activation, which is in source.avg.mom (or in source.trial(i).mom). The three rows of the moment are the strength of the source in the x-, y- and z-direction. Using ft_sourcedescriptives with the option cfg.projectmom=yes you can project the source to its strongest orientation, i.e. the direction that explains most of the source variance. That is equivalent to taking the largest eigenvector of the source timeseries (which is a phrasing that is often used in papers, also on combining fMRI bold timeseries over multiple voxels). You can also do it manually, like this mom3d = randn(3,100); % example source dipole moment [u, s, v] = svd(mom, 'econ'); mom1d = v(1,:); Alternative to taking the largest eigenvector (which has a sign-ambiguity), you can do something like mom3d = randn(3,100); % example source dipole moment [u, s, v] = svd(mom, 'econ'); mom1d = abs(v(1,:)); to take the absolute value, or mom3d = randn(3,100); % example source dipole moment mom1d = sqrt(sum(mom3d.^2,1)); to take the strength over all three directions for each timepoint. best Robert On 25 Aug 2011, at 19:43, Rojas, Don wrote: > Tony, > > I don't use the LCMV beamformer, so I can't tell you what the structure of the output is, but what you need to perform a source space projection is the orientation and location information for the sources. In the DICS beamformer output, which I'm familiar with, this information would be in the source structure as follows (assuming your source output structure is called "source"): > > source.pos % the position info as an n x 3 matrix, where n is the number of sources, in the same units as your volume conductor model > source.avg.ori % the orientation as an 1 x n cell array, where n is the number of locations and each cell is a 3 x 1 array of x y z orientation and n refers to the % same source in each array > > So, in that case, you might find the necessary information for the peak source, in my case, as follows: > > [~,ind]=max(source.avg.pow(:)); % index of the source with the maximum power in the dics output, let's say ind = 5000 > ori=source.avg.ori(5000); % orientation info > loc=source.pos(5000,:); % location information > > Then, if you get this info, you can search the archives for my posts, one of which has a detailed explanation for getting the source waveform when you have a known source location, source orientation and a volume conductor. If you want to get all the waveforms, you can of course iterate the process over the entire voxel volume, but since the leadfields are highly correlated in adjacent source locations, you will see that many of the waveforms will look essentially the same for sources within a few centimeters. > > Best, > > Don > ----------------------- > Don Rojas, Ph.D. > Associate Professor of Psychiatry > U. of Colorado Denver Anschutz Medical Campus Director, UCD > Magnetoencephalography Lab > 13001 E. 17th Pl F546 > Aurora, CO 80045 USA > 303-724-4994 > > On Aug 23, 2011, at 12:50 PM, Passaro, Antony D wrote: > >> Hi Fieldtrippers, >> >> I was looking for a way to extract a time-course from each source estimated by the lcmv beamformer. I searched through the mailing list archives and came across a few emails which talked about the possibility but I was unable to find any examples explaining exactly how this process is done. I am interested in extracting the time-course over the average response rather than the individual trials, if possible. Any help would be greatly appreciated. >> >> Thanks, >> -Tony >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From irina.simanova at mpi.nl Tue Aug 30 08:03:29 2011 From: irina.simanova at mpi.nl (Irina Simanova) Date: Tue, 30 Aug 2011 08:03:29 +0200 Subject: [FieldTrip] ICA on MEG data Message-ID: <729C68C8-BB67-44F7-B439-A5B9C90BBB9D@mpi.nl> Dear FieldTripers, I would like to identify ocular artifacts in an MEG dataset using the ICA. The data consist of 180 trials of variable duration ( the signal from stimulus onset to subject's response). The trials' duration vary from 400 ms to 3000 ms. The data is sampled at 1.2 kHz, and has 275 channels. I am new to ICA, and I need a piece of advice. What would be a reasonable number of output independent components? Should I do the PCA as a pre-processing stage? How do i find an optimal number of principal components to extract? I would appreciate your help, Kind regards, Irina Simanova PhD student Neurobiology of Language Group Max-Planck Institute for Psycholinguistics Wundtlaan 1 6525 XD Nijmegen The Netherlands e-mail: irina.simanova at mpi.nl phone: +31 24 3521541 From r.oostenveld at donders.ru.nl Tue Aug 30 09:08:19 2011 From: r.oostenveld at donders.ru.nl (Robert Oostenveld) Date: Tue, 30 Aug 2011 09:08:19 +0200 Subject: [FieldTrip] Source Timecourse In-Reply-To: References: <51AF637A-B1EB-4421-84B3-6B0D0C4BBA24@ucdenver.edu> Message-ID: <9370AD38-AA89-4038-957B-28525CAA0A37@donders.ru.nl> Hi Tony On 29 Aug 2011, at 18:33, Passaro, Antony D wrote: > Thank you both for your feedback, I really appreciate it. Using the ft_sourcedescriptives function (and projectmom), I was able to extract the time-course for each source. I am still curious about the output though as it appears to represent the entire duration (including the baseline) of the epoch rather than the specified time-window defined during timelockanalysis (just prior to sourceanalysis). Default behaviour is to construct the covariance and average for the whole timewindow, indeed including the baseline period. Depending on the experimental data, the LCMV filter can be optimized for a specific timewindow by computing the covariance only for that timewindow. Subsequently the filter can be applied to the whole timewindow (including baseline) or only to the timewindow of interest. > That being the case, does this time-course represent the .pow estimated by the lcmv beamformer or is the nai? source.avg.pow is the power estimated based on the data covariance window, not on the longer time window that you might pass through the filter. If you computed the covariance for a smaller timewindow (using ft_timelockanalysis), then only that is used for the power estimate. The same applies to the nai. > Or is it simply a source representation of the gradiometers over the entire epoch? Another issue I came across concerns the source interpolation. It seems once I interpolate the sources using a template mri, only the pow and nai fields are interpolated....is there a way to also interpolate the mom field as well or is it a matter of backtracking to figure out which original source corresponds to each interpolated source? The mom timecourse is indeed not interpolated, as that used to exceed the memory on many computers: the interpolated mom would consist of a full 3D volume at each timepoint of your ERF. We are in the process of making some improvements to the source data structure, i.e. to its representation. The planning is mostly done, but implementation wise I don't know how far the code already is. I think you could work with the following to interpolate it source.avg.momint = nan(length(source.pos,1), length(source.time)); for i=1:length(source.inside) indx = source.inside(i); source.avg.momint(indx,:) = source.avg.mom{indx}; % the grid locations outside the brain keep their NaN end and then specify cfg.parameter = 'momint' in ft_sourceinterpolate. best Robert From stan.vanpelt at fcdonders.ru.nl Tue Aug 30 09:14:31 2011 From: stan.vanpelt at fcdonders.ru.nl (Stan van Pelt) Date: Tue, 30 Aug 2011 09:14:31 +0200 (CEST) Subject: [FieldTrip] ICA on MEG data In-Reply-To: <729C68C8-BB67-44F7-B439-A5B9C90BBB9D@mpi.nl> References: <729C68C8-BB67-44F7-B439-A5B9C90BBB9D@mpi.nl> Message-ID: <0db501cc66e4$76b60a00$64221e00$@vanpelt@fcdonders.ru.nl> Dear Irina, There's a page on the Fieldtrip Wiki about using ICA to remove EOG-artifacts, see http://fieldtrip.fcdonders.nl/example/use_independent_component_analysis_i ca_to_remove_eog_artifacts. The common approach is to first define your trials, and then run ICA on them. You will get as many output components as you have channels, i.e. 275 in your case. The EOG artifacts usually pop up in 1 or 2 components within the first 20 largest components. Best regards, Stan Stan van Pelt, PhD Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen Kapittelweg 29, 6525 EN Nijmegen, Netherlands E-mail: stan.vanpelt at donders.ru.nl Website: www.ru.nl/donders/ Phone: (+31) (0)24 36 68495 Fax: (+31) (0)24 36 10989 -----Original Message----- From: fieldtrip-bounces at donders.ru.nl [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Irina Simanova Sent: Tuesday, August 30, 2011 8:03 AM To: Email discussion list for the FieldTrip project Subject: [FieldTrip] ICA on MEG data Dear FieldTripers, I would like to identify ocular artifacts in an MEG dataset using the ICA. The data consist of 180 trials of variable duration ( the signal from stimulus onset to subject's response). The trials' duration vary from 400 ms to 3000 ms. The data is sampled at 1.2 kHz, and has 275 channels. I am new to ICA, and I need a piece of advice. What would be a reasonable number of output independent components? Should I do the PCA as a pre-processing stage? How do i find an optimal number of principal components to extract? I would appreciate your help, Kind regards, Irina Simanova PhD student Neurobiology of Language Group Max-Planck Institute for Psycholinguistics Wundtlaan 1 6525 XD Nijmegen The Netherlands e-mail: irina.simanova at mpi.nl phone: +31 24 3521541 _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From irina.simanova at mpi.nl Tue Aug 30 09:56:12 2011 From: irina.simanova at mpi.nl (Irina Simanova) Date: Tue, 30 Aug 2011 09:56:12 +0200 Subject: [FieldTrip] ICA on MEG data In-Reply-To: <0db501cc66e4$76b60a00$64221e00$@vanpelt@fcdonders.ru.nl> References: <729C68C8-BB67-44F7-B439-A5B9C90BBB9D@mpi.nl> <0db501cc66e4$76b60a00$64221e00$@vanpelt@fcdonders.ru.nl> Message-ID: <79A92C58-4FA2-4ADD-B6D6-975A4C4B9F08@mpi.nl> Dear Stan, Thank you for your reply, As far as I understand from the tutorials about ICA by Scott Makeig http://sccn.ucsd.edu/~scott/tutorial/icafaq.html , the number of time points should be considerably larger than the number of scalp sensors. How critical is this point? Besides, the sensors are very close to each other, and the registered activity in adjacent sensors is highly correlated, so the components would be similar, right? Isn't it reasonable to reduce the number of output components then? Thank you in advance, Irina On Aug 30, 2011, at 9:14 AM 8/30/11, Stan van Pelt wrote: > Dear Irina, > > There's a page on the Fieldtrip Wiki about using ICA to remove > EOG-artifacts, see > http://fieldtrip.fcdonders.nl/example/use_independent_component_analysis_i > ca_to_remove_eog_artifacts. > > The common approach is to first define your trials, and then run ICA > on > them. > > You will get as many output components as you have channels, i.e. > 275 in > your case. The EOG artifacts usually pop up in 1 or 2 components > within > the first 20 largest components. > > Best regards, > Stan > > Stan van Pelt, PhD > Donders Institute for Brain, Cognition and Behaviour, Radboud > University > Nijmegen > Kapittelweg 29, 6525 EN Nijmegen, Netherlands > E-mail: stan.vanpelt at donders.ru.nl > Website: www.ru.nl/donders/ > Phone: (+31) (0)24 36 68495 > Fax: (+31) (0)24 36 10989 > > -----Original Message----- > From: fieldtrip-bounces at donders.ru.nl > [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Irina Simanova > Sent: Tuesday, August 30, 2011 8:03 AM > To: Email discussion list for the FieldTrip project > Subject: [FieldTrip] ICA on MEG data > > Dear FieldTripers, > > I would like to identify ocular artifacts in an MEG dataset using the > ICA. > > The data consist of 180 trials of variable duration ( the signal from > stimulus onset to subject's response). The trials' duration vary from > 400 ms to 3000 ms. The data is sampled at 1.2 kHz, and has 275 > channels. I am new to ICA, and I need a piece of advice. What would be > a reasonable number of output independent components? > Should I do the PCA as a pre-processing stage? How do i find an > optimal number of principal components to extract? > > I would appreciate your help, > > Kind regards, > > Irina Simanova > PhD student > Neurobiology of Language Group > Max-Planck Institute for Psycholinguistics > Wundtlaan 1 > 6525 XD Nijmegen > The Netherlands > e-mail: irina.simanova at mpi.nl > phone: +31 24 3521541 > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From stan.vanpelt at fcdonders.ru.nl Tue Aug 30 10:15:07 2011 From: stan.vanpelt at fcdonders.ru.nl (Stan van Pelt) Date: Tue, 30 Aug 2011 10:15:07 +0200 (CEST) Subject: [FieldTrip] ICA on MEG data In-Reply-To: <79A92C58-4FA2-4ADD-B6D6-975A4C4B9F08@mpi.nl> References: <729C68C8-BB67-44F7-B439-A5B9C90BBB9D@mpi.nl> <0db501cc66e4$76b60a00$64221e00$@vanpelt@fcdonders.ru.nl> <79A92C58-4FA2-4ADD-B6D6-975A4C4B9F08@mpi.nl> Message-ID: <0e0601cc66ec$ee2048f0$ca60dad0$@vanpelt@fcdonders.ru.nl> Dear Irina, The number of time points should indeed be much larger for a good ICA estimate. However, that seems to be the case with your data, so I think there's no problem there. As far as I know, with ICA in fieldtrip you always get the same amount of output components as input channels. Activity in adjacent MEG channels will be somewhat correlated indeed, but not nearly as much as with EEG. Moreover, the origin of your artifacts (EOG, maybe also ECG) will create a distinct topographic distribution, and the components will pop out clearly from your ft_componentanalysis, because it explains large amounts of variance in the data. Just give it a try, I'd say. You should get similar results as in the Wiki example. Best, Stan -----Original Message----- From: fieldtrip-bounces at donders.ru.nl [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Irina Simanova Sent: Tuesday, August 30, 2011 9:56 AM To: Email discussion list for the FieldTrip project Subject: Re: [FieldTrip] ICA on MEG data Dear Stan, Thank you for your reply, As far as I understand from the tutorials about ICA by Scott Makeig http://sccn.ucsd.edu/~scott/tutorial/icafaq.html , the number of time points should be considerably larger than the number of scalp sensors. How critical is this point? Besides, the sensors are very close to each other, and the registered activity in adjacent sensors is highly correlated, so the components would be similar, right? Isn't it reasonable to reduce the number of output components then? Thank you in advance, Irina On Aug 30, 2011, at 9:14 AM 8/30/11, Stan van Pelt wrote: > Dear Irina, > > There's a page on the Fieldtrip Wiki about using ICA to remove > EOG-artifacts, see > http://fieldtrip.fcdonders.nl/example/use_independent_component_analysis_i > ca_to_remove_eog_artifacts. > > The common approach is to first define your trials, and then run ICA > on > them. > > You will get as many output components as you have channels, i.e. > 275 in > your case. The EOG artifacts usually pop up in 1 or 2 components > within > the first 20 largest components. > > Best regards, > Stan > > Stan van Pelt, PhD > Donders Institute for Brain, Cognition and Behaviour, Radboud > University > Nijmegen > Kapittelweg 29, 6525 EN Nijmegen, Netherlands > E-mail: stan.vanpelt at donders.ru.nl > Website: www.ru.nl/donders/ > Phone: (+31) (0)24 36 68495 > Fax: (+31) (0)24 36 10989 > > -----Original Message----- > From: fieldtrip-bounces at donders.ru.nl > [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Irina Simanova > Sent: Tuesday, August 30, 2011 8:03 AM > To: Email discussion list for the FieldTrip project > Subject: [FieldTrip] ICA on MEG data > > Dear FieldTripers, > > I would like to identify ocular artifacts in an MEG dataset using the > ICA. > > The data consist of 180 trials of variable duration ( the signal from > stimulus onset to subject's response). The trials' duration vary from > 400 ms to 3000 ms. The data is sampled at 1.2 kHz, and has 275 > channels. I am new to ICA, and I need a piece of advice. What would be > a reasonable number of output independent components? > Should I do the PCA as a pre-processing stage? How do i find an > optimal number of principal components to extract? > > I would appreciate your help, > > Kind regards, > > Irina Simanova > PhD student > Neurobiology of Language Group > Max-Planck Institute for Psycholinguistics > Wundtlaan 1 > 6525 XD Nijmegen > The Netherlands > e-mail: irina.simanova at mpi.nl > phone: +31 24 3521541 > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From k.muesch at uke.uni-hamburg.de Tue Aug 30 10:44:42 2011 From: k.muesch at uke.uni-hamburg.de (=?iso-8859-1?Q?Kathrin_M=FCsch?=) Date: Tue, 30 Aug 2011 10:44:42 +0200 Subject: [FieldTrip] ICA on MEG data In-Reply-To: <0e0601cc66ec$ee2048f0$ca60dad0$@vanpelt@fcdonders.ru.nl> References: <729C68C8-BB67-44F7-B439-A5B9C90BBB9D@mpi.nl> <0db501cc66e4$76b60a00$64221e00$@vanpelt@fcdonders.ru.nl> <79A92C58-4FA2-4ADD-B6D6-975A4C4B9F08@mpi.nl> <0e0601cc66ec$ee2048f0$ca60dad0$@vanpelt@fcdonders.ru.nl> Message-ID: Dear Irina, I reduced the number of components to 64 by running a PCA beforehand (cfg.runica.pca = 64) on my MEG data. Depending on how much the subjects moved their eyes, blinks and saccades popped up in earlier or later components. I had feasible results with this approach both for EOG and ECG. Good luck, Kathrin _____________________________________ Kathrin Müsch Dept. of Neurophysiology and Pathophysiology University Medical Center Hamburg-Eppendorf Martinistr. 52 20246 Hamburg Germany Phone: +49-40-7410-54680 Fax: +49-40-7410-57752 E-Mail: k.muesch at uke.uni-hamburg.de _____________________________________ Am 30.08.2011 um 10:15 schrieb Stan van Pelt: > Dear Irina, > > The number of time points should indeed be much larger for a good ICA > estimate. However, that seems to be the case with your data, so I think > there's no problem there. > > As far as I know, with ICA in fieldtrip you always get the same amount of > output components as input channels. Activity in adjacent MEG channels > will be somewhat correlated indeed, but not nearly as much as with EEG. > Moreover, the origin of your artifacts (EOG, maybe also ECG) will create a > distinct topographic distribution, and the components will pop out clearly > from your ft_componentanalysis, because it explains large amounts of > variance in the data. > > Just give it a try, I'd say. You should get similar results as in the Wiki > example. > > Best, > Stan > > > -----Original Message----- > From: fieldtrip-bounces at donders.ru.nl > [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Irina Simanova > Sent: Tuesday, August 30, 2011 9:56 AM > To: Email discussion list for the FieldTrip project > Subject: Re: [FieldTrip] ICA on MEG data > > Dear Stan, > > Thank you for your reply, > > As far as I understand from the tutorials about ICA by Scott Makeig > http://sccn.ucsd.edu/~scott/tutorial/icafaq.html > , the number of time points should be considerably larger than the > number of scalp sensors. How critical is this point? > > Besides, the sensors are very close to each other, and the registered > activity in adjacent sensors is highly correlated, so the components > would be similar, right? Isn't it reasonable to reduce the number of > output components then? > > Thank you in advance, > Irina > > > > On Aug 30, 2011, at 9:14 AM 8/30/11, Stan van Pelt wrote: > >> Dear Irina, >> >> There's a page on the Fieldtrip Wiki about using ICA to remove >> EOG-artifacts, see >> > http://fieldtrip.fcdonders.nl/example/use_independent_component_analysis_i >> ca_to_remove_eog_artifacts. >> >> The common approach is to first define your trials, and then run ICA >> on >> them. >> >> You will get as many output components as you have channels, i.e. >> 275 in >> your case. The EOG artifacts usually pop up in 1 or 2 components >> within >> the first 20 largest components. >> >> Best regards, >> Stan >> >> Stan van Pelt, PhD >> Donders Institute for Brain, Cognition and Behaviour, Radboud >> University >> Nijmegen >> Kapittelweg 29, 6525 EN Nijmegen, Netherlands >> E-mail: stan.vanpelt at donders.ru.nl >> Website: www.ru.nl/donders/ >> Phone: (+31) (0)24 36 68495 >> Fax: (+31) (0)24 36 10989 >> >> -----Original Message----- >> From: fieldtrip-bounces at donders.ru.nl >> [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Irina Simanova >> Sent: Tuesday, August 30, 2011 8:03 AM >> To: Email discussion list for the FieldTrip project >> Subject: [FieldTrip] ICA on MEG data >> >> Dear FieldTripers, >> >> I would like to identify ocular artifacts in an MEG dataset using the >> ICA. >> >> The data consist of 180 trials of variable duration ( the signal from >> stimulus onset to subject's response). The trials' duration vary from >> 400 ms to 3000 ms. The data is sampled at 1.2 kHz, and has 275 >> channels. I am new to ICA, and I need a piece of advice. What would be >> a reasonable number of output independent components? >> Should I do the PCA as a pre-processing stage? How do i find an >> optimal number of principal components to extract? >> >> I would appreciate your help, >> >> Kind regards, >> >> Irina Simanova >> PhD student >> Neurobiology of Language Group >> Max-Planck Institute for Psycholinguistics >> Wundtlaan 1 >> 6525 XD Nijmegen >> The Netherlands >> e-mail: irina.simanova at mpi.nl >> phone: +31 24 3521541 >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From irina.simanova at mpi.nl Tue Aug 30 14:22:31 2011 From: irina.simanova at mpi.nl (Irina Simanova) Date: Tue, 30 Aug 2011 14:22:31 +0200 Subject: [FieldTrip] Call for Papers :: NIPS 2011 Workshop on Interpretable Decoding of Higher Cognitive States from Neural Data Message-ID: <2E7CF2CC-C7AF-4937-935E-D0337E8445C8@mpi.nl> Interpretable Decoding of Higher Cognitive States from Neural Data NIPS 2011 Workshop, Dec 16 or 17, 2011, Granada, Spain (Please feel free to distribute the CFP to all the interested persons and groups.) Overview Over recent years, machine learning methods have become a crucial analytical tool in cognitive neuroscience (see reviews by Formisano et al., 2008; Pereira et al., 2009). Decoding techniques have dramatically increased the sensitivity of experiments, and so also the subtlety of cognitive questions that can be asked. At the same time the mental phenomena being studied are moving beyond lower-level perceptual and motor processes which are directly grounded in external measurable realities. Decoding higher cognition and interpreting the learned behaviour of the classifiers used pose unique challenges, as these psychological states are complex, fast-changing and often ill-defined. Contemporary machine learning methods deal well with the small numbers of cases, and high numbers of co-linear dimensions typical of neural data, and are generally optimized to maximize classification performance, rather than to enable meaningful interpretation of the features they learn from. And indeed recent work has succeeded to decode psychological phenomena including visual object recognition (e.g. Kriegeskorte et al., 2008; Connolly et al., 2011), perceptual interpretation of sounds (Staeren et al., 2009), lexical semantics (Mitchell et al., 2008; Siminova et al., 2010; Devereux et al., 2010; Murphy et al., 2011), decision making during game playing (Xiang et al., 2009) and the process of mental arithmetic (Anderson et al., 2008). But for the cognitive scientists who use these methods, the primary question is often not "how much" but rather "how" and "why" the patterns of neural activity identified by a machine learning algorithm encode particular cognitive processes. The aim of this workshop is therefore to 1) discuss the achievements and problems of the decoding of high-level cognitive states, and 2) explore the use of machine learning methodologies and other computational models that enable such cognitive interpretation of neural recordings of different modalities. Advances in this field require close collaboration between machine learning experts, neuroscientists and cognitive scientists. Thus, this workshop is highly interdisciplinary and will aim to attract submissions also from outside the existing NIPS community. By stimulating discussions among experts in the different fields, the workshop seeks to generate novel insights and new directions for research. Topics of interest The field requires techniques that are capable of taking advantage of spatially distributed patterns in the brain, that are separated in space but coordinated in their activity. Methods should also be sensitive to the fine-grained temporal patterns of multiple processes - which may proceed in a serial fashion, overlapping or in parallel with each-other, or in multiple passes with bidirectional information flows. Different recording modalities have distinctive advantages: fMRI provides very fine millimetre-level localisation in the brain but poor temporal resolution, while EEG and MEG have millisecond temporal resolution at the cost of spatial resolution. Ideally machine learning methods would be able to meaningfully combine complementary information from these different neuroimaging techniques (see e.g. De Martino et al., 2010). Moreover, as the processes underlying higher cognition are so complex, methods should be able to disentangle even tightly linked and confounded subprocesses. Finally, general use algorithms that could induce latent dimensions from neural data, and so reveal the "hidden" psychological states, would be a dramatic advance on current hypothesis-driven analytical paradigms. Originality of approach is encouraged and submissions on any related methodological approach are welcomed, such as: - Interpreting spatial and temporal location of selected features and their weights - Discovering "hidden" or "latent" cognitive representations - Disentangling confounded processes and representations - Comparing or combining data from recording modalities (e.g. fMRI, EEG, structural MRI, DTI, MEG, NIRS, EcOG, single cell recordings) - Fuzzy and partial classifications - Unaligned or incommensurate feature spaces and data representation As noted above, the complexity of higher cognition poses challenges. To take language comprehension as an example, speech is received at 3-4 words; acoustic, semantic and syntactic processing can occur in parallel; and the form of underlying representations (sentence structures, conceptual descriptions) remains controversial. We welcome submissions dealing with any high-level cognitive functions that exhibit similar complexity, for instance: - Knowledge representation and concepts - Language and communication - Understanding visual and auditory experience - Memory and learning - Reasoning and problem solving - Decision making and executive control Submissions Authors are invited to submit full papers on original, unpublished work in the topic area of this workshop via the NIPS 2011 submission site at https://cmt.research.microsoft.com/NIPS2011/Default.aspx. Submissions should be formatted using the NIPS 2011 stylefiles, with blind review and not exceeding 8 pages plus an extra page for references. Author and submission information can be found at http://nips.cc/PaperInformation/AuthorSubmissionInstructions . The stylefiles are available at http://nips.cc/PaperInformation/StyleFiles . Each submission will be reviewed at least by two members of the programme committee. Accepted papers will be published in the workshop proceedings. Dual submissions to the main NIPS 2011 conference and this workshop are allowed; if you submit to the main session, indicate this when you submit to the workshop. If your paper is accepted for the main session, you should withdraw your paper from the workshop upon notification by the main session. Important Dates - Aug 30, 2011: Call for papers - Sep 23, 2011: Deadline for submission of workshop papers - Oct 15, 2011: Notification of acceptance - Oct 31, 2011: Camera-ready papers due - Dec 16 or 17, 2011: Workshop date Links - NIPS 2011 website: http://nips.cc/Conferences/2011/ - Workshop website: https://sites.google.com/site/decodehighcogstate - Call for Papers: https://sites.google.com/site/decodehighcogstate/cfp/ Kind regards, The Workshop Organizers, Kai-min Kevin Chang, Anna Korhonen, Irina Simanova, Brian Murphy -------------- next part -------------- An HTML attachment was scrubbed... URL: From anna.lambrechts at gmail.com Tue Aug 30 17:46:52 2011 From: anna.lambrechts at gmail.com (Anna Lambrechts) Date: Tue, 30 Aug 2011 17:46:52 +0200 Subject: [FieldTrip] Issues with neighbours in ft_timelockstatistics Message-ID: Dear fieldtrip community, Until about two weeks ago I was using a batch to process group comparison statistics on MEG data using ft_timelockstatistics. A week later, the same batch was not working anymore, even on the same data. The error I get is the following: *??? Improper index matrix reference. Error in ==> clusterstat>makechanneighbstructmat at 520 [seld] = match_str(cfg.channel, cfg.neighbours(chan).label); Error in ==> clusterstat at 60 channeighbstructmat = makechanneighbstructmat(cfg); Error in ==> statistics_montecarlo at 320 [stat, cfg] = clusterstat(cfg, statrand, statobs,'issource',issource); Error in ==> statistics_wrapper at 290 [stat, cfg] = statmethod(cfg, dat, design, 'issource',issource); Error in ==> ft_timelockstatistics at 124 [stat, cfg] = statistics_wrapper(cfg, varargin{:}); Error in ==> myfunction_stats at 59 gpstat = ft_timelockstatistics(cfg, data1, data2); *It seems the error concerns the 'neighbours.mat' file reading, even though I am not sure about it. When I delete the two last lines in this file (to match the size of matrix wanted) the error disappears, but the processing is incomplete and false (or at least it does not match the results I had before). Has anyone encounter the same issue? Did you do any update recently that might explain this problem, and do you have any idea how to remedy to it? Best wishes, Anna. -------------- next part -------------- An HTML attachment was scrubbed... URL: From ekanal at cmu.edu Tue Aug 30 17:58:34 2011 From: ekanal at cmu.edu (Kanal Eliezer) Date: Tue, 30 Aug 2011 11:58:34 -0400 Subject: [FieldTrip] assessing significance in using ft_timelockanalysis results In-Reply-To: <2109_1314385277_p7QJ1Gjs003376_020701cc6422$65b10f30$31132d90$@maris@donders.ru.nl> References: <8FC0FE65-8E0D-4D7C-9A62-CE1610845C92@cmu.edu> <2109_1314385277_p7QJ1Gjs003376_020701cc6422$65b10f30$31132d90$@maris@donders.ru.nl> Message-ID: <8CC6BA25-9A03-4630-B940-486BE2B9EE92@cmu.edu> Hello Eric - Thanks for the response. It looks like the fixed vs random analysis is exactly what I'm referring to. From what I understood, it looks like the difference really only shows up in the variance of the resultant distribution; with a random, the variance also takes into account the betwee-subjects variance. Is there a way to specify whether I want to do a fixed or random effects analysis in FieldTrip when I'm running ft_timelockgrandaverage or ft_freqgrandaverage? Thanks! Elli p.s. - In case anyone else is trying to figure this out, chapter 12 of Friston's book "Statistical Parametric Mapping" does an excellent job explaining the difference between fixed and random analyses, as well as how to implement it algorithmically. On Aug 26, 2011, at 3:00 PM, Eric Maris wrote: > Hi Kanal, > >> The event related statistics tutorial >> (http://fieldtrip.fcdonders.nl/tutorial/eventrelatedstatistics) talks > about >> assessing significance parametrically by running t-tests on pooled >> timelockanalysis data. My question is, does the fact that the averages > were >> created from N trials make a difference? If I'm condition A has twelve >> averages and condition B has another twelve, and each average contains 70 >> trials, is there a way to "inform" the statistical test that the power in > this >> dataset is greater than 24? Is this only possible if I run the t-test > comparing >> each set of 840 (70*12) trials? >> >> I'm also curious whether this is possible with non-parametric analyses, as >> well. Thanks - > > In an analysis over subjects (called random-effects analysis in the fMRI > literature), "informing" the statistical test about the number of trials per > condition only makes sense if this number is different for the two > conditions. I propose that you have a look the fMRI papers that deal with > the issue of fixed-versus-random effect analyses. The conceptual issues > involved are the same in fMRI and electrophysiology. > > > Best, > > Eric Maris > > > > >> >> Elli Kanal >> >> >> -------------------- >> Eliezer Kanal, Ph.D. >> Postdoctoral Fellow >> Center for the Neural Basis of Cognition >> Carnegie Mellon University >> 4400 Fifth Ave, Suite 110A >> Pittsburgh PA 15213 >> P: 412-268-4115 >> F: 412-268-5060 >> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From e.maris at donders.ru.nl Tue Aug 30 21:13:42 2011 From: e.maris at donders.ru.nl (Eric Maris) Date: Tue, 30 Aug 2011 21:13:42 +0200 Subject: [FieldTrip] assessing significance in using ft_timelockanalysis results Message-ID: <00d201cc6748$f18e8fe0$d4abafa0$@maris@donders.ru.nl> Hi Elli, > Thanks for the response. It looks like the fixed vs random analysis is exactly > what I'm referring to. From what I understood, it looks like the difference > really only shows up in the variance of the resultant distribution; with a > random, the variance also takes into account the betwee-subjects variance. > Is there a way to specify whether I want to do a fixed or random effects > analysis in FieldTrip when I'm running ft_timelockgrandaverage or > ft_freqgrandaverage? Thanks! I don't get this. Ft_timelockgrandaverage and ft_freqgrandaverage will only be called when a random effects analysis is performed. Best, Eric > > Elli > > p.s. - In case anyone else is trying to figure this out, chapter 12 of Friston's > book "Statistical Parametric Mapping" does an excellent job explaining the > difference between fixed and random analyses, as well as how to implement > it algorithmically. > > > > On Aug 26, 2011, at 3:00 PM, Eric Maris wrote: > > > Hi Kanal, > > > >> The event related statistics tutorial > >> (http://fieldtrip.fcdonders.nl/tutorial/eventrelatedstatistics) talks > > about > >> assessing significance parametrically by running t-tests on pooled > >> timelockanalysis data. My question is, does the fact that the averages > > were > >> created from N trials make a difference? If I'm condition A has twelve > >> averages and condition B has another twelve, and each average contains > 70 > >> trials, is there a way to "inform" the statistical test that the power in > > this > >> dataset is greater than 24? Is this only possible if I run the t-test > > comparing > >> each set of 840 (70*12) trials? > >> > >> I'm also curious whether this is possible with non-parametric analyses, as > >> well. Thanks - > > > > In an analysis over subjects (called random-effects analysis in the fMRI > > literature), "informing" the statistical test about the number of trials per > > condition only makes sense if this number is different for the two > > conditions. I propose that you have a look the fMRI papers that deal with > > the issue of fixed-versus-random effect analyses. The conceptual issues > > involved are the same in fMRI and electrophysiology. > > > > > > Best, > > > > Eric Maris > > > > > > > > > >> > >> Elli Kanal > >> > >> > >> -------------------- > >> Eliezer Kanal, Ph.D. > >> Postdoctoral Fellow > >> Center for the Neural Basis of Cognition > >> Carnegie Mellon University > >> 4400 Fifth Ave, Suite 110A > >> Pittsburgh PA 15213 > >> P: 412-268-4115 > >> F: 412-268-5060 > >> > >> > >> _______________________________________________ > >> fieldtrip mailing list > >> fieldtrip at donders.ru.nl > >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > _______________________________________________ > > fieldtrip mailing list > > fieldtrip at donders.ru.nl > > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From ekanal at cmu.edu Tue Aug 30 21:46:24 2011 From: ekanal at cmu.edu (Kanal Eliezer) Date: Tue, 30 Aug 2011 15:46:24 -0400 Subject: [FieldTrip] assessing significance in using ft_timelockanalysis results In-Reply-To: <29602_1314731679_p7UJEcWB023970_00d201cc6748$f18e8fe0$d4abafa0$@maris@donders.ru.nl> References: <29602_1314731679_p7UJEcWB023970_00d201cc6748$f18e8fe0$d4abafa0$@maris@donders.ru.nl> Message-ID: <4D99A829-9BB1-4395-9997-45568224DAEC@cmu.edu> Maybe I had been using it incorrectly, then. In my study, I have numerous datafiles for each subject (each block gets split into it's own .fif file, due to file size), and I use those functions to combine the preprocessed and timelocked datafiles. Should I be doing this differently? Elli On Aug 30, 2011, at 3:13 PM, Eric Maris wrote: > Hi Elli, > > > >> Thanks for the response. It looks like the fixed vs random analysis is > exactly >> what I'm referring to. From what I understood, it looks like the > difference >> really only shows up in the variance of the resultant distribution; with a >> random, the variance also takes into account the betwee-subjects variance. >> Is there a way to specify whether I want to do a fixed or random effects >> analysis in FieldTrip when I'm running ft_timelockgrandaverage or >> ft_freqgrandaverage? Thanks! > > > I don't get this. Ft_timelockgrandaverage and ft_freqgrandaverage will only > be called when a random effects analysis is performed. > > > Best, > > Eric > > >> >> Elli >> >> p.s. - In case anyone else is trying to figure this out, chapter 12 of > Friston's >> book "Statistical Parametric Mapping" does an excellent job explaining the >> difference between fixed and random analyses, as well as how to implement >> it algorithmically. >> >> >> >> On Aug 26, 2011, at 3:00 PM, Eric Maris wrote: >> >>> Hi Kanal, >>> >>>> The event related statistics tutorial >>>> (http://fieldtrip.fcdonders.nl/tutorial/eventrelatedstatistics) talks >>> about >>>> assessing significance parametrically by running t-tests on pooled >>>> timelockanalysis data. My question is, does the fact that the averages >>> were >>>> created from N trials make a difference? If I'm condition A has twelve >>>> averages and condition B has another twelve, and each average contains >> 70 >>>> trials, is there a way to "inform" the statistical test that the power > in >>> this >>>> dataset is greater than 24? Is this only possible if I run the t-test >>> comparing >>>> each set of 840 (70*12) trials? >>>> >>>> I'm also curious whether this is possible with non-parametric analyses, > as >>>> well. Thanks - >>> >>> In an analysis over subjects (called random-effects analysis in the fMRI >>> literature), "informing" the statistical test about the number of trials > per >>> condition only makes sense if this number is different for the two >>> conditions. I propose that you have a look the fMRI papers that deal > with >>> the issue of fixed-versus-random effect analyses. The conceptual issues >>> involved are the same in fMRI and electrophysiology. >>> >>> >>> Best, >>> >>> Eric Maris >>> >>> >>> >>> >>>> >>>> Elli Kanal >>>> >>>> >>>> -------------------- >>>> Eliezer Kanal, Ph.D. >>>> Postdoctoral Fellow >>>> Center for the Neural Basis of Cognition >>>> Carnegie Mellon University >>>> 4400 Fifth Ave, Suite 110A >>>> Pittsburgh PA 15213 >>>> P: 412-268-4115 >>>> F: 412-268-5060 >>>> >>>> >>>> _______________________________________________ >>>> fieldtrip mailing list >>>> fieldtrip at donders.ru.nl >>>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >>> >>> _______________________________________________ >>> fieldtrip mailing list >>> fieldtrip at donders.ru.nl >>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From semrich at wisc.edu Tue Aug 30 21:49:03 2011 From: semrich at wisc.edu (Stephen Emrich) Date: Tue, 30 Aug 2011 14:49:03 -0500 Subject: [FieldTrip] depsamplesregrT help Message-ID: Dear fieldtrip users and developers, I am attempting to use the depsamplesregrT statistic, but have encountered some problems. The documentation also appears to be incomplete. I am attempting to regress the outputs of ft_freqanalysis for 3 conditions against some values (the IVs). From what I can tell, depsamplesregrT should do this. Some issues I'm running into: cvar: cvar is not documented in this function, but i'm assuming from some searches that the cvar should be another row of values that actually correspond to the regressors (the IVs). If I attempt to run it without specifying cvar, I get an error. unitselved: even if I include the cvar in the design matrix, there is an error that 'unitselved' is undefined. I can't find any documentation on what this variable is or what the values should be. Any help would be appreciated. S- From jonas at obleser.de Wed Aug 31 12:10:40 2011 From: jonas at obleser.de (Jonas Obleser) Date: Wed, 31 Aug 2011 12:10:40 +0200 Subject: [FieldTrip] depsamplesregrT help Message-ID: <67EC1721-BF53-450B-A1DB-67F16F4E2CC7@obleser.de> Hi Stephen, from my understanding (greatly facilitated by Eric Maris a year ago or so), depsamplesregrT does only test for linear increases or decreases in your data, depending on the sequence in which you input your data conditions. For example, freqstatistics(cfg, cond1{:}, cond2{:}, cond3{:}) will result in positive clusters for data increasing (quasi-)linearly from cond1 to cond3, and negative clusters for data behaving vice versa. For regression/correlation with an independent (e.g., behavioural) regressor, we use a statfun Nathan Weisz (also active on the list here) once wrote and kindly shared with us. Hope this helps somewhat. Best wishes, J. Dr. Jonas Obleser Auditory Cognition Group Max Planck Institute for Human Cognitive and Brain Sciences Leipzig, Germany (p) +49 (0)341 9940 114 (e) obleser at cbs.mpg.de From jm.horschig at donders.ru.nl Wed Aug 31 13:02:52 2011 From: jm.horschig at donders.ru.nl (=?ISO-8859-1?Q?=22J=F6rn_M=2E_Horschig=22?=) Date: Wed, 31 Aug 2011 13:02:52 +0200 Subject: [FieldTrip] Issues with neighbours in ft_timelockstatistics In-Reply-To: References: Message-ID: <4E5E14DC.8050000@donders.ru.nl> Hi Anna, As you might have read, we recently updated how neighbours are computed and stored, and that you have to specify your neighbours manually now. Could you let me know how you defined the neighbours and how many channels your data has? Maybe there is a mismatch between the neighbour struct and your data in number of sensors, but that's just a wild guess. Anyhow, the result might look different depending on how you define your neighbours. Anyway, if you provide more information I can look into this and see how to fix it. Best, Jörn On 8/30/2011 5:46 PM, Anna Lambrechts wrote: > Dear fieldtrip community, > > Until about two weeks ago I was using a batch to process group > comparison statistics on MEG data using ft_timelockstatistics. A week > later, the same batch was not working anymore, even on the same data. > > The error I get is the following: > > *??? Improper index matrix reference. > > Error in ==> clusterstat>makechanneighbstructmat at 520 > [seld] = match_str(cfg.channel, cfg.neighbours(chan).label); > > Error in ==> clusterstat at 60 > channeighbstructmat = makechanneighbstructmat(cfg); > > Error in ==> statistics_montecarlo at 320 > [stat, cfg] = clusterstat(cfg, statrand, statobs,'issource',issource); > > Error in ==> statistics_wrapper at 290 > [stat, cfg] = statmethod(cfg, dat, design, 'issource',issource); > > Error in ==> ft_timelockstatistics at 124 > [stat, cfg] = statistics_wrapper(cfg, varargin{:}); > > Error in ==> myfunction_stats at 59 > gpstat = ft_timelockstatistics(cfg, data1, data2); > > *It seems the error concerns the 'neighbours.mat' file reading, even > though I am not sure about it. When I delete the two last lines in > this file (to match the size of matrix wanted) the error disappears, > but the processing is incomplete and false (or at least it does not > match the results I had before). > > Has anyone encounter the same issue? > Did you do any update recently that might explain this problem, and do > you have any idea how to remedy to it? > > Best wishes, > Anna. > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip -- Jörn M. Horschig PhD Student Donders Institute for Brain, Cognition and Behaviour Centre for Cognitive Neuroimaging Radboud University Nijmegen Neuronal Oscillations Group P.O. Box 9101 NL-6500 HB Nijmegen The Netherlands Contact: E-Mail: jm.horschig at donders.ru.nl Tel: +31-(0)24-36-68493 Web: http://www.ru.nl/donders Visiting address: Trigon, room 2.30 Kapittelweg 29 NL-6525 EN Nijmegen The Netherlands -------------- next part -------------- An HTML attachment was scrubbed... URL: From anna.lambrechts at gmail.com Wed Aug 31 13:40:33 2011 From: anna.lambrechts at gmail.com (Anna Lambrechts) Date: Wed, 31 Aug 2011 13:40:33 +0200 Subject: [FieldTrip] fieldtrip Digest, Vol 9, Issue 30 In-Reply-To: References: Message-ID: Hello, I have written yesterday about troubles while running ft_timelockstatistics. As it happens now, this seems to be a compatibility issue between matlab 2011 and the current fieldtrip version (which you can solve temporarily by using older versions of both). Best wishes, Anna. 2011/8/30 > Send fieldtrip mailing list submissions to > fieldtrip at donders.ru.nl > > To subscribe or unsubscribe via the World Wide Web, visit > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > or, via email, send a message with subject or body 'help' to > fieldtrip-request at donders.ru.nl > > You can reach the person managing the list at > fieldtrip-owner at donders.ru.nl > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of fieldtrip digest..." > > > Today's Topics: > > 1. Call for Papers :: NIPS 2011 Workshop on Interpretable > Decoding of Higher Cognitive States from Neural Data (Irina Simanova) > 2. Issues with neighbours in ft_timelockstatistics (Anna Lambrechts) > 3. Re: assessing significance in using ft_timelockanalysis > results (Kanal Eliezer) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Tue, 30 Aug 2011 14:22:31 +0200 > From: Irina Simanova > To: Email discussion list for the FieldTrip project > > Subject: [FieldTrip] Call for Papers :: NIPS 2011 Workshop on > Interpretable Decoding of Higher Cognitive States from Neural Data > Message-ID: <2E7CF2CC-C7AF-4937-935E-D0337E8445C8 at mpi.nl> > Content-Type: text/plain; charset="us-ascii"; Format="flowed"; > DelSp="yes" > > Interpretable Decoding of Higher Cognitive States from Neural Data > > NIPS 2011 Workshop, Dec 16 or 17, 2011, Granada, Spain > > (Please feel free to distribute the CFP to all the interested persons > and groups.) > > Overview > > Over recent years, machine learning methods have become a crucial > analytical tool in cognitive neuroscience (see reviews by Formisano et > al., 2008; Pereira et al., 2009). Decoding techniques have > dramatically increased the sensitivity of experiments, and so also the > subtlety of cognitive questions that can be asked. At the same time > the mental phenomena being studied are moving beyond lower-level > perceptual and motor processes which are directly grounded in external > measurable realities. > > Decoding higher cognition and interpreting the learned behaviour of > the classifiers used pose unique challenges, as these psychological > states are complex, fast-changing and often ill-defined. Contemporary > machine learning methods deal well with the small numbers of cases, > and high numbers of co-linear dimensions typical of neural data, and > are generally optimized to maximize classification performance, rather > than to enable meaningful interpretation of the features they learn > from. And indeed recent work has succeeded to decode psychological > phenomena including visual object recognition (e.g. Kriegeskorte et > al., 2008; Connolly et al., 2011), perceptual interpretation of sounds > (Staeren et al., 2009), lexical semantics (Mitchell et al., 2008; > Siminova et al., 2010; Devereux et al., 2010; Murphy et al., 2011), > decision making during game playing (Xiang et al., 2009) and the > process of mental arithmetic (Anderson et al., 2008). But for the > cognitive scientists who use these methods, the primary question is > often not "how much" but rather "how" and "why" the patterns of neural > activity identified by a machine learning algorithm encode particular > cognitive processes. > > The aim of this workshop is therefore to 1) discuss the achievements > and problems of the decoding of high-level cognitive states, and 2) > explore the use of machine learning methodologies and other > computational models that enable such cognitive interpretation of > neural recordings of different modalities. Advances in this field > require close collaboration between machine learning experts, > neuroscientists and cognitive scientists. Thus, this workshop is > highly interdisciplinary and will aim to attract submissions also from > outside the existing NIPS community. By stimulating discussions among > experts in the different fields, the workshop seeks to generate novel > insights and new directions for research. > > Topics of interest > > The field requires techniques that are capable of taking advantage of > spatially distributed patterns in the brain, that are separated in > space but coordinated in their activity. Methods should also be > sensitive to the fine-grained temporal patterns of multiple processes > - which may proceed in a serial fashion, overlapping or in parallel > with each-other, or in multiple passes with bidirectional information > flows. Different recording modalities have distinctive advantages: > fMRI provides very fine millimetre-level localisation in the brain but > poor temporal resolution, while EEG and MEG have millisecond temporal > resolution at the cost of spatial resolution. Ideally machine learning > methods would be able to meaningfully combine complementary > information from these different neuroimaging techniques (see e.g. De > Martino et al., 2010). Moreover, as the processes underlying higher > cognition are so complex, methods should be able to disentangle even > tightly linked and confounded subprocesses. Finally, general use > algorithms that could induce latent dimensions from neural data, and > so reveal the "hidden" psychological states, would be a dramatic > advance on current hypothesis-driven analytical paradigms. Originality > of approach is encouraged and submissions on any related > methodological approach are welcomed, such as: > > - Interpreting spatial and temporal location of selected features and > their weights > - Discovering "hidden" or "latent" cognitive representations > - Disentangling confounded processes and representations > - Comparing or combining data from recording modalities (e.g. fMRI, > EEG, structural MRI, DTI, MEG, NIRS, EcOG, single cell recordings) > - Fuzzy and partial classifications > - Unaligned or incommensurate feature spaces and data representation > > As noted above, the complexity of higher cognition poses challenges. > To take language comprehension as an example, speech is received at > 3-4 words; acoustic, semantic and syntactic processing can occur in > parallel; and the form of underlying representations (sentence > structures, conceptual descriptions) remains controversial. We welcome > submissions dealing with any high-level cognitive functions that > exhibit similar complexity, for instance: > > - Knowledge representation and concepts > - Language and communication > - Understanding visual and auditory experience > - Memory and learning > - Reasoning and problem solving > - Decision making and executive control > > Submissions > > Authors are invited to submit full papers on original, unpublished > work in the topic area of this workshop via the NIPS 2011 submission > site at https://cmt.research.microsoft.com/NIPS2011/Default.aspx. > Submissions should be formatted using the NIPS 2011 stylefiles, with > blind review and not exceeding 8 pages plus an extra page for > references. Author and submission information can be found at > http://nips.cc/PaperInformation/AuthorSubmissionInstructions > . The stylefiles are available at > http://nips.cc/PaperInformation/StyleFiles > . Each submission will be reviewed at least by two members of the > programme committee. Accepted papers will be published in the workshop > proceedings. Dual submissions to the main NIPS 2011 conference and > this workshop are allowed; if you submit to the main session, indicate > this when you submit to the workshop. If your paper is accepted for > the main session, you should withdraw your paper from the workshop > upon notification by the main session. > > Important Dates > > - Aug 30, 2011: Call for papers > - Sep 23, 2011: Deadline for submission of workshop papers > - Oct 15, 2011: Notification of acceptance > - Oct 31, 2011: Camera-ready papers due > - Dec 16 or 17, 2011: Workshop date > > Links > > - NIPS 2011 website: http://nips.cc/Conferences/2011/ > - Workshop website: https://sites.google.com/site/decodehighcogstate > - Call for Papers: https://sites.google.com/site/decodehighcogstate/cfp/ > > Kind regards, > The Workshop Organizers, > Kai-min Kevin Chang, Anna Korhonen, Irina Simanova, Brian Murphy > > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: < > http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20110830/27b062df/attachment-0001.html > > > > ------------------------------ > > Message: 2 > Date: Tue, 30 Aug 2011 17:46:52 +0200 > From: Anna Lambrechts > To: fieldtrip at donders.ru.nl > Subject: [FieldTrip] Issues with neighbours in ft_timelockstatistics > Message-ID: > > > Content-Type: text/plain; charset="iso-8859-1" > > Dear fieldtrip community, > > Until about two weeks ago I was using a batch to process group comparison > statistics on MEG data using ft_timelockstatistics. A week later, the same > batch was not working anymore, even on the same data. > > The error I get is the following: > > *??? Improper index matrix reference. > > Error in ==> clusterstat>makechanneighbstructmat at 520 > [seld] = match_str(cfg.channel, cfg.neighbours(chan).label); > > Error in ==> clusterstat at 60 > channeighbstructmat = makechanneighbstructmat(cfg); > > Error in ==> statistics_montecarlo at 320 > [stat, cfg] = clusterstat(cfg, statrand, statobs,'issource',issource); > > Error in ==> statistics_wrapper at 290 > [stat, cfg] = statmethod(cfg, dat, design, 'issource',issource); > > Error in ==> ft_timelockstatistics at 124 > [stat, cfg] = statistics_wrapper(cfg, varargin{:}); > > Error in ==> myfunction_stats at 59 > gpstat = ft_timelockstatistics(cfg, data1, data2); > > *It seems the error concerns the 'neighbours.mat' file reading, even though > I am not sure about it. When I delete the two last lines in this file (to > match the size of matrix wanted) the error disappears, but the processing > is > incomplete and false (or at least it does not match the results I had > before). > > Has anyone encounter the same issue? > Did you do any update recently that might explain this problem, and do you > have any idea how to remedy to it? > > Best wishes, > Anna. > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: < > http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20110830/f81c8aad/attachment-0001.html > > > > ------------------------------ > > Message: 3 > Date: Tue, 30 Aug 2011 11:58:34 -0400 > From: Kanal Eliezer > To: Email discussion list for the FieldTrip project > > Subject: Re: [FieldTrip] assessing significance in using > ft_timelockanalysis results > Message-ID: <8CC6BA25-9A03-4630-B940-486BE2B9EE92 at cmu.edu> > Content-Type: text/plain; charset=us-ascii > > Hello Eric - > > Thanks for the response. It looks like the fixed vs random analysis is > exactly what I'm referring to. From what I understood, it looks like the > difference really only shows up in the variance of the resultant > distribution; with a random, the variance also takes into account the > betwee-subjects variance. Is there a way to specify whether I want to do a > fixed or random effects analysis in FieldTrip when I'm running > ft_timelockgrandaverage or ft_freqgrandaverage? Thanks! > > Elli > > p.s. - In case anyone else is trying to figure this out, chapter 12 of > Friston's book "Statistical Parametric Mapping" does an excellent job > explaining the difference between fixed and random analyses, as well as how > to implement it algorithmically. > > > > On Aug 26, 2011, at 3:00 PM, Eric Maris wrote: > > > Hi Kanal, > > > >> The event related statistics tutorial > >> (http://fieldtrip.fcdonders.nl/tutorial/eventrelatedstatistics) talks > > about > >> assessing significance parametrically by running t-tests on pooled > >> timelockanalysis data. My question is, does the fact that the averages > > were > >> created from N trials make a difference? If I'm condition A has twelve > >> averages and condition B has another twelve, and each average contains > 70 > >> trials, is there a way to "inform" the statistical test that the power > in > > this > >> dataset is greater than 24? Is this only possible if I run the t-test > > comparing > >> each set of 840 (70*12) trials? > >> > >> I'm also curious whether this is possible with non-parametric analyses, > as > >> well. Thanks - > > > > In an analysis over subjects (called random-effects analysis in the fMRI > > literature), "informing" the statistical test about the number of trials > per > > condition only makes sense if this number is different for the two > > conditions. I propose that you have a look the fMRI papers that deal with > > the issue of fixed-versus-random effect analyses. The conceptual issues > > involved are the same in fMRI and electrophysiology. > > > > > > Best, > > > > Eric Maris > > > > > > > > > >> > >> Elli Kanal > >> > >> > >> -------------------- > >> Eliezer Kanal, Ph.D. > >> Postdoctoral Fellow > >> Center for the Neural Basis of Cognition > >> Carnegie Mellon University > >> 4400 Fifth Ave, Suite 110A > >> Pittsburgh PA 15213 > >> P: 412-268-4115 > >> F: 412-268-5060 > >> > >> > >> _______________________________________________ > >> fieldtrip mailing list > >> fieldtrip at donders.ru.nl > >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > _______________________________________________ > > fieldtrip mailing list > > fieldtrip at donders.ru.nl > > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > > ------------------------------ > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > End of fieldtrip Digest, Vol 9, Issue 30 > **************************************** > -------------- next part -------------- An HTML attachment was scrubbed... URL: From amelie.serpollet at cea.fr Wed Aug 3 10:30:38 2011 From: amelie.serpollet at cea.fr (=?iso-8859-1?Q?SERPOLLET_Am=E9lie_228173?=) Date: Wed, 3 Aug 2011 08:30:38 +0000 Subject: [FieldTrip] delay with realtime buffer Message-ID: <04B8DFAA8CFBED46A48C995E5BB1CD459A39@EXDAG0-B0.intra.cea.fr> Dear Fieldtrip users and developers, Using Fieldtrip buffer for a BCI loop (in realtime), I want to measure the delay introduced by the use of the buffer. I use two simple programs to measure it : the first one sends data to a buffer and writes CPU time in a file when a rising edge was detected in the signal sent ; the other program requests the data and also writes CPU time in a file when the same rising edge is detected in the signal received. I checked that the delay introduced by the detection and file writing is insignificant (0.5 to 5 µs). On my computer, with 32 channels signal and chunks of 8 to 32 points, I measured very variables delays, also after deactivating the anti-virus : 1ms to more than 45ms. Do you know what it is due to, if I can reduce it or if I can make it less variable ? Thank you in advance. Amélie -------------- next part -------------- An HTML attachment was scrubbed... URL: From Don.Rojas at ucdenver.edu Wed Aug 3 17:47:56 2011 From: Don.Rojas at ucdenver.edu (Rojas, Don) Date: Wed, 3 Aug 2011 09:47:56 -0600 Subject: [FieldTrip] ft_volumesegment question Message-ID: <621FD28E-EB95-4966-9612-B64308F79074@ucdenver.edu> Dear Fieldtrippers, I use FieldTrip and SPM8 together quite frequently. I noticed recently that the segmentation output (gray, white and csf tpm) of ft_volumesegment (c1, c2, c3) does not align properly to the input mri scan. This is not the case when doing the same segmentation in SPM8. I'm using the 20110725 version of Fieldtrip and have mad certain that it and not the one in spm8 is used. My cfg and command is as follows: cfg.output = {'brain' 'skull' 'scalp','tpm'}; cfg.spmversion = 'spm8'; cfg.name = 'test'; cfg.write = 'yes'; cfg.coordsys = 'spm'; seg = ft_volumesegment(cfg,mri); Looking at the code for ft_volumesegment, it appears that a permutation of the mri dimensions (swapping 2 and 3) is done on line 279: [mri,permutevec,flipflags] = align_ijk2xyz(mri); The mri.transform field is changed accordingly. However, the original.transform is put into the headers for the c1, c2 and c3 output (line 399). I think that this is not correct because the segmentation output has the permuted dimensions, not the dimensions of the original mri. So, the correct transform should be the mri.transform after align_ijk2xyz I think. Or maybe this is intentional behavior and I'm missing something. It can easily be checked, however, using Check Reg, in SPM8, that the output tpm do not align correctly with the original mri. If I manually write the mri.transform (from align_ijk2xyz) to the tpm output, however, they do align correctly. Best, Don ----------------------- Don Rojas, Ph.D. Associate Professor of Psychiatry U. of Colorado Denver Anschutz Medical Campus Director, UCD Magnetoencephalography Lab 13001 E. 17th Pl F546 Aurora, CO 80045 303-724-4994 -------------- next part -------------- An HTML attachment was scrubbed... URL: From ekanal at cmu.edu Wed Aug 3 23:29:52 2011 From: ekanal at cmu.edu (Kanal Eliezer) Date: Wed, 3 Aug 2011 17:29:52 -0400 Subject: [FieldTrip] timelock grand average can't average gradiometers? Message-ID: <2788E298-FE53-4F74-9C27-FDC29C0DDCBF@cmu.edu> Hello all - I'm trying to average a number of datasets collected from a Neuromag scanner. This dataset contains both magnetometer and gradiometer data. I'm trying to use ft_timelockgrandaverage to average the cleaned data, and I'm getting the following error message: > Warning: discarding gradiometer information because it cannot be averaged My question is, why not? My gradiometer data isn't different than the magnetometer data. Even more confusing, when looking in the code, it looks like it's averaging it anyway, and the resulting dataset seems to contain averaged grad data. What's the purpose of this error? Thanks, Eliezer Kanal From eelke.spaak at donders.ru.nl Thu Aug 4 08:03:32 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Thu, 4 Aug 2011 08:03:32 +0200 Subject: [FieldTrip] timelock grand average can't average gradiometers? In-Reply-To: <2788E298-FE53-4F74-9C27-FDC29C0DDCBF@cmu.edu> References: <2788E298-FE53-4F74-9C27-FDC29C0DDCBF@cmu.edu> Message-ID: Hi Eli, The warning message you mention regards the information about the gradiometer positions, i.e. the information in the data.grad field. The MEG-data recorded by the gradiometer sensors will still be averaged. I understand the confusion, though, and agree that the message could be a bit clearer. Best, Eelke 2011/8/3 Kanal Eliezer : > Hello all - > > I'm trying to average a number of datasets collected from a Neuromag scanner. This dataset contains both magnetometer and gradiometer data. I'm trying to use ft_timelockgrandaverage to average the cleaned data, and I'm getting the following error message: > >> Warning: discarding gradiometer information because it cannot be averaged > > My question is, why not? My gradiometer data isn't different than the magnetometer data. Even more confusing, when looking in the code, it looks like it's averaging it anyway, and the resulting dataset seems to contain averaged grad data. What's the purpose of this error? > > Thanks, > Eliezer Kanal > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From anna.lambrechts at gmail.com Thu Aug 4 11:17:46 2011 From: anna.lambrechts at gmail.com (Anna Lambrechts) Date: Thu, 4 Aug 2011 11:17:46 +0200 Subject: [FieldTrip] fieldtrip Digest, Vol 8, Issue 40 In-Reply-To: References: Message-ID: Hi, thank you for your first answer. Actually my question is really on the right way to implement a group comparison in fieldtrip (patient vs control group). Is it ok to run an independent t-test with the following design: design = zeros(2,n_group1+n_group2); design(2,:) = 1; design(1,1:n_group1) = 1:n_group1; design(1,n_group1+1:end) = 1:n_group2; A subsequent question is whether it is possible to implement a mixed design or we always have to proceed by combining conditions and compare it between groups. Cheers, Anna. > Message: 2 > Date: Sat, 30 Jul 2011 20:45:12 +0200 > From: "Eric Maris" > To: "'Email discussion list for the FieldTrip project'" > > Subject: Re: [FieldTrip] Group comparison statistics - ERF study > Message-ID: <03a601cc4ee8$d152b760$73f82620$@maris at donders.ru.nl> > Content-Type: text/plain; charset="us-ascii" > > Hi Anna, > > > > You can compare the two groups with respect to any linear combination that > can be formed using the observations in the 2 (conditions) x 2 (response > types) within-UO design. For testing interaction effects, these linear > combinations are the usual contrasts. > > > > > > Best, > > > > Eric Maris > > > > > > dr. Eric Maris > Donders Institute for Brain, Cognition and Behavior > > Radboud University > P.O. Box 9104 > 6500 HE Nijmegen > The Netherlands > T:+31 24 3612651 > Mobile: 06 39584581 > > F:+31 24 3616066 > mailto:e.maris at donders.ru.nl > > http://www.nphyscog.com/ > > > > > > > > From: fieldtrip-bounces at donders.ru.nl > [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Anna Lambrechts > Sent: donderdag 28 juli 2011 11:32 > To: fieldtrip at donders.ru.nl > Subject: [FieldTrip] Group comparison statistics - ERF study > > > > Hi, > > I am trying to run a group comparison analysis on event-related fields data > in a 2 (groups) x 2 (conditions) x 2 (response types) design. Is this > possible at all with any fieldtrip script? As far as I know implemented > statistics look at within-groups comparison. > > Thanks, > Anna. > > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: < > http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20110730/1f410391/attachment-0001.html > > > > ------------------------------ > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > End of fieldtrip Digest, Vol 8, Issue 40 > **************************************** > -------------- next part -------------- An HTML attachment was scrubbed... URL: From v.litvak at ion.ucl.ac.uk Thu Aug 4 12:39:53 2011 From: v.litvak at ion.ucl.ac.uk (Vladimir Litvak) Date: Thu, 4 Aug 2011 11:39:53 +0100 Subject: [FieldTrip] timelock grand average can't average gradiometers? In-Reply-To: References: <2788E298-FE53-4F74-9C27-FDC29C0DDCBF@cmu.edu> Message-ID: Dear Eelke, Actually there is a way to average sensors which is implemented in ft_average_sens. Perhaps it would be a good idea to use it in ft_timelockgrandaverage. Best, Vladimir On Thu, Aug 4, 2011 at 7:03 AM, Eelke Spaak wrote: > Hi Eli, > > The warning message you mention regards the information about the > gradiometer positions, i.e. the information in the data.grad field. > The MEG-data recorded by the gradiometer sensors will still be > averaged. > > I understand the confusion, though, and agree that the message could > be a bit clearer. > > Best, > Eelke > > 2011/8/3 Kanal Eliezer : >> Hello all - >> >> I'm trying to average a number of datasets collected from a Neuromag scanner. This dataset contains both magnetometer and gradiometer data. I'm trying to use ft_timelockgrandaverage to average the cleaned data, and I'm getting the following error message: >> >>> Warning: discarding gradiometer information because it cannot be averaged >> >> My question is, why not? My gradiometer data isn't different than the magnetometer data. Even more confusing, when looking in the code, it looks like it's averaging it anyway, and the resulting dataset seems to contain averaged grad data. What's the purpose of this error? >> >> Thanks, >> Eliezer Kanal >> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From eelke.spaak at donders.ru.nl Thu Aug 4 14:04:44 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Thu, 4 Aug 2011 14:04:44 +0200 Subject: [FieldTrip] timelock grand average can't average gradiometers? In-Reply-To: References: <2788E298-FE53-4F74-9C27-FDC29C0DDCBF@cmu.edu> Message-ID: Dear Vladimir, Thanks, I did not know about that function. However, I don't think ft_timelockgrandaverage should automatically average sensor positions; if the user wants this it is probably best if he/she does this explicitly. But, I will bring it up in the next FieldTrip dev team meeting, and see what others think (particularly Robert and Jan-Mathijs). Best, Eelke 2011/8/4 Vladimir Litvak : > Dear Eelke, > > Actually there is a way to average sensors which is implemented in > ft_average_sens. Perhaps it would be a good idea to use it in > ft_timelockgrandaverage. > > Best, > > Vladimir > > On Thu, Aug 4, 2011 at 7:03 AM, Eelke Spaak wrote: >> Hi Eli, >> >> The warning message you mention regards the information about the >> gradiometer positions, i.e. the information in the data.grad field. >> The MEG-data recorded by the gradiometer sensors will still be >> averaged. >> >> I understand the confusion, though, and agree that the message could >> be a bit clearer. >> >> Best, >> Eelke >> >> 2011/8/3 Kanal Eliezer : >>> Hello all - >>> >>> I'm trying to average a number of datasets collected from a Neuromag scanner. This dataset contains both magnetometer and gradiometer data. I'm trying to use ft_timelockgrandaverage to average the cleaned data, and I'm getting the following error message: >>> >>>> Warning: discarding gradiometer information because it cannot be averaged >>> >>> My question is, why not? My gradiometer data isn't different than the magnetometer data. Even more confusing, when looking in the code, it looks like it's averaging it anyway, and the resulting dataset seems to contain averaged grad data. What's the purpose of this error? >>> >>> Thanks, >>> Eliezer Kanal >>> >>> >>> _______________________________________________ >>> fieldtrip mailing list >>> fieldtrip at donders.ru.nl >>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >>> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From e.maris at donders.ru.nl Thu Aug 4 14:08:39 2011 From: e.maris at donders.ru.nl (Eric Maris) Date: Thu, 4 Aug 2011 14:08:39 +0200 Subject: [FieldTrip] fieldtrip Digest, Vol 8, Issue 40 In-Reply-To: References: Message-ID: <053f01cc529f$3efff020$bcffd060$@maris@donders.ru.nl> Hi Anna, thank you for your first answer. Actually my question is really on the right way to implement a group comparison in fieldtrip (patient vs control group). Is it ok to run an independent t-test with the following design: design = zeros(2,n_group1+n_group2); design(2,:) = 1; design(1,1:n_group1) = 1:n_group1; design(1,n_group1+1:end) = 1:n_group2; This is not correct. I advise you to have a look at the Fieldtrip tutorial on cluster-based permutation tests. A between-subjects experiment (e.g., patients versus controls) is formally equivalent to a between-trials experiment (explained in detail in the tutorial), and therefore should be analyzed in the same way. You can even copy-paste part of the tutorial code for the analysis of your between-subjects data. A subsequent question is whether it is possible to implement a mixed design or we always have to proceed by combining conditions and compare it between groups. I do not see the contrast between these two options. What is wrong with the second option? In fact, it is the option that is advocated in many statistics books (e.g., Maxwell & Delaney, Johnson & Wichren). Best, Eric Maris Cheers, Anna. Message: 2 Date: Sat, 30 Jul 2011 20:45:12 +0200 From: "Eric Maris" To: "'Email discussion list for the FieldTrip project'" Subject: Re: [FieldTrip] Group comparison statistics - ERF study Message-ID: <03a601cc4ee8$d152b760$73f82620$@maris at donders.ru.nl> Content-Type: text/plain; charset="us-ascii" Hi Anna, You can compare the two groups with respect to any linear combination that can be formed using the observations in the 2 (conditions) x 2 (response types) within-UO design. For testing interaction effects, these linear combinations are the usual contrasts. Best, Eric Maris dr. Eric Maris Donders Institute for Brain, Cognition and Behavior Radboud University P.O. Box 9104 6500 HE Nijmegen The Netherlands T:+31 24 3612651 Mobile: 06 39584581 F:+31 24 3616066 mailto:e.maris at donders.ru.nl http://www.nphyscog.com/ From: fieldtrip-bounces at donders.ru.nl [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Anna Lambrechts Sent: donderdag 28 juli 2011 11:32 To: fieldtrip at donders.ru.nl Subject: [FieldTrip] Group comparison statistics - ERF study Hi, I am trying to run a group comparison analysis on event-related fields data in a 2 (groups) x 2 (conditions) x 2 (response types) design. Is this possible at all with any fieldtrip script? As far as I know implemented statistics look at within-groups comparison. Thanks, Anna. -------------- next part -------------- An HTML attachment was scrubbed... URL: ------------------------------ _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip End of fieldtrip Digest, Vol 8, Issue 40 **************************************** -------------- next part -------------- An HTML attachment was scrubbed... URL: From drivolta81 at gmail.com Thu Aug 4 14:44:08 2011 From: drivolta81 at gmail.com (Davide Rivolta) Date: Thu, 4 Aug 2011 14:44:08 +0200 Subject: [FieldTrip] Potential bug in trl Message-ID: Dear all, I have noticed that the trial definition works well when begsample, ensample and offset have fixed values. I have however a specific trialfunction, where each trial has a different lenght. This works well when you specify a fixed offset, however, when even the offset has to change for each trial, Fieldtrip reports values until 0. It ignores time after 0. Might it be a bug? Thanks, Davide -------------- next part -------------- An HTML attachment was scrubbed... URL: From BelluscB at ninds.nih.gov Thu Aug 4 15:23:57 2011 From: BelluscB at ninds.nih.gov (Belluscio, Beth (NIH/NINDS) [E]) Date: Thu, 4 Aug 2011 09:23:57 -0400 Subject: [FieldTrip] time frequency analysis - basic question Message-ID: <794DFDD3C128EF44AC49ADC58FD0090C057456B617@NIHMLBX10.nih.gov> Hi all- When calling ft_freqanalysis you specify the foi limits and intervals - eg 5:5:30. The resulting data set appears to generate the power at discreet frequencies (5, 10, 15, 20, 25, 30). Does this actually reflect an estimate of the power over a range of frequencies (2.5-7.5, 7.5-12.5, 12.5-17.5,....) or the power at discreet frequencies with those in between not evaluated? Thanks for your help, Beth. Beth Belluscio MD-PhD Clinical Fellow Human Motor Control Section NINDS, NIH 301-402-3495 -------------- next part -------------- An HTML attachment was scrubbed... URL: From michael.wibral at web.de Thu Aug 4 15:55:46 2011 From: michael.wibral at web.de (Michael Wibral) Date: Thu, 4 Aug 2011 15:55:46 +0200 (CEST) Subject: [FieldTrip] Potential bug in trl Message-ID: <422581888.3319195.1312466146646.JavaMail.fmail@mwmweb086> An HTML attachment was scrubbed... URL: From drivolta81 at gmail.com Thu Aug 4 16:03:24 2011 From: drivolta81 at gmail.com (Davide Rivolta) Date: Thu, 4 Aug 2011 16:03:24 +0200 Subject: [FieldTrip] Potential bug in trl In-Reply-To: <422581888.3319195.1312466146646.JavaMail.fmail@mwmweb086> References: <422581888.3319195.1312466146646.JavaMail.fmail@mwmweb086> Message-ID: Thanks Michael, I wrote my own trialfun. The offset would be negative and would have the same lenght of difference between endsample-begsample. This, for each trial, would have a different value. It can be a bug. In fact, I tried to vhange the trialfun by leaving the same begsample and endsample, but I put a fixed offset (i.e., -100) and it worked. All the trial was in and not, like with the variable offset, until time = 0. Thanks, Davide On Thu, Aug 4, 2011 at 3:55 PM, Michael Wibral wrote: > Hi davide, > > I guess in such a case you would have to write your own trial function. > However, offset values larger than 0 mean, that the stimulus occured before > you start your piece of data - is that really what you wanted ? > > Michael > > > ------------------------------ > *Von:* "Davide Rivolta" > *Gesendet:* Aug 4, 2011 2:44:08 PM > *An:* fieldtrip at donders.ru.nl > *Betreff:* [FieldTrip] Potential bug in trl > > > Dear all, > > I have noticed that the trial definition works well when begsample, > ensample and offset have fixed values. > > I have however a specific trialfunction, where each trial has a different > lenght. > > This works well when you specify a fixed offset, however, when even the > offset has to change for each trial, Fieldtrip reports values until 0. It > ignores time after 0. > > > Might it be a bug? > > Thanks, > > Davide > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > -- Davide Rivolta, PhD -------------- next part -------------- An HTML attachment was scrubbed... URL: From jm.horschig at donders.ru.nl Fri Aug 5 09:13:23 2011 From: jm.horschig at donders.ru.nl (=?ISO-8859-1?Q?=22J=F6rn_M=2E_Horschig=22?=) Date: Fri, 05 Aug 2011 09:13:23 +0200 Subject: [FieldTrip] Potential bug in trl In-Reply-To: References: <422581888.3319195.1312466146646.JavaMail.fmail@mwmweb086> Message-ID: <4E3B9813.8070507@donders.ru.nl> Dear Davide, I never worked with trials of different length, however this should be straight forward when writing your own trialfun (just as Michael proposed). In most trialfuns, the trl matrix is constructed by concatenating the vectors begsample, endsample and offset. These vectors can have any values you want them to have. As a simple example, you can write a trialfun that just returns trl = [1 2 1; 400 500 100] (for your experiment, you of course want these values to be computed from your trigger values). Is your problem that such a thing does not work? It would be great if you give a concrete use-case, or e.g. attach your trialfun. Best, Jörn On 8/4/2011 4:03 PM, Davide Rivolta wrote: > Thanks Michael, > I wrote my own trialfun. > The offset would be negative and would have the same lenght of > difference between endsample-begsample. This, for each trial, would > have a different value. > It can be a bug. In fact, I tried to vhange the trialfun by leaving > the same begsample and endsample, but I put a fixed offset (i.e., > -100) and it worked. All the trial was in and not, like with the > variable offset, until time = 0. > Thanks, > Davide > > On Thu, Aug 4, 2011 at 3:55 PM, Michael Wibral > wrote: > > Hi davide, > > I guess in such a case you would have to write your own trial > function. However, offset values larger than 0 mean, that the > stimulus occured before you start your piece of data - is that > really what you wanted ? > > Michael > > > ------------------------------------------------------------------------ > *Von:* "Davide Rivolta" > > *Gesendet:* Aug 4, 2011 2:44:08 PM > *An:* fieldtrip at donders.ru.nl > *Betreff:* [FieldTrip] Potential bug in trl > > > Dear all, > I have noticed that the trial definition works well when > begsample, ensample and offset have fixed values. > I have however a specific trialfunction, where each trial has > a different lenght. > This works well when you specify a fixed offset, however, when > even the offset has to change for each trial, Fieldtrip > reports values until 0. It ignores time after 0. > Might it be a bug? > Thanks, > Davide > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > > -- > Davide Rivolta, PhD > > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip -- Jörn M. Horschig PhD Student Donders Institute for Brain, Cognition and Behaviour Centre for Cognitive Neuroimaging Radboud University Nijmegen Neuronal Oscillations Group P.O. Box 9101 NL-6500 HB Nijmegen The Netherlands Contact: E-Mail: jm.horschig at donders.ru.nl Tel: +31-(0)24-36-68493 Web: http://www.ru.nl/donders Visiting address: Trigon, room 2.30 Kapittelweg 29 NL-6525 EN Nijmegen The Netherlands -------------- next part -------------- An HTML attachment was scrubbed... URL: From jm.horschig at donders.ru.nl Fri Aug 5 09:29:38 2011 From: jm.horschig at donders.ru.nl (=?ISO-8859-1?Q?=22J=F6rn_M=2E_Horschig=22?=) Date: Fri, 05 Aug 2011 09:29:38 +0200 Subject: [FieldTrip] time frequency analysis - basic question In-Reply-To: <794DFDD3C128EF44AC49ADC58FD0090C057456B617@NIHMLBX10.nih.gov> References: <794DFDD3C128EF44AC49ADC58FD0090C057456B617@NIHMLBX10.nih.gov> Message-ID: <4E3B9BE2.4000300@donders.ru.nl> Dear Beth, From all my knowledge I gathered so far, I can try to answer your question (and hope that people from this list correct me if I am wrong): As often, the truth lies somewhere in between. The frequency smoothing is determined by the type and length of the taper you apply - this is similar to wavelet analysis. If you use multitapers, you can specify cfg.tapsmofrq, which determines the frequency band you want to smooth with. The FT freqanalysis method uses multiplication in the time domain instead of convolution in the frequency domain (which is equivalent). Any taper will always have a specific frequency spectrum itself, and thus it will not give you the value of a single frequency (though the output looks as if). So if you specify foi as [5 10 15], then the power at 5Hz will not be from 2.5 to 7.5Hz, but neither will it be at exactly and only 5Hz (btw, the exact center frequency might not be 5Hz, this depends on the (Rayleigh) frequency resolution of your signal). I do not know any more specifics concerning this, maybe the help of ft_freqanalysis can tell you some additonal things: http://fieldtrip.fcdonders.nl/reference/ft_freqanalysis Best, Jörn Btw, if this is not correct, else from the mailinglist should correct me asap ;) On 8/4/2011 3:23 PM, Belluscio, Beth (NIH/NINDS) [E] wrote: > > Hi all- > > When calling ft_freqanalysis you specify the foi limits and > intervals -- eg 5:5:30. The resulting data set appears to generate > the power at discreet frequencies (5, 10, 15, 20, 25, 30). Does this > actually reflect an estimate of the power over a range of frequencies > (2.5-7.5, 7.5-12.5, 12.5-17.5,....) or the power at discreet > frequencies with those in between not evaluated? > > Thanks for your help, > > Beth. > > Beth Belluscio MD-PhD > > Clinical Fellow > > Human Motor Control Section > > NINDS, NIH > > 301-402-3495 > > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip -- Jörn M. Horschig PhD Student Donders Institute for Brain, Cognition and Behaviour Centre for Cognitive Neuroimaging Radboud University Nijmegen Neuronal Oscillations Group P.O. Box 9101 NL-6500 HB Nijmegen The Netherlands Contact: E-Mail: jm.horschig at donders.ru.nl Tel: +31-(0)24-36-68493 Web: http://www.ru.nl/donders Visiting address: Trigon, room 2.30 Kapittelweg 29 NL-6525 EN Nijmegen The Netherlands -------------- next part -------------- An HTML attachment was scrubbed... URL: From drivolta81 at gmail.com Fri Aug 5 11:36:07 2011 From: drivolta81 at gmail.com (Davide Rivolta) Date: Fri, 5 Aug 2011 11:36:07 +0200 Subject: [FieldTrip] Potential bug in trl In-Reply-To: <4E3B9813.8070507@donders.ru.nl> References: <422581888.3319195.1312466146646.JavaMail.fmail@mwmweb086> <4E3B9813.8070507@donders.ru.nl> Message-ID: Dear Jörn, Thank you for your reply. I guess I have now figured it out. Thanks, Davide On Fri, Aug 5, 2011 at 9:13 AM, "Jörn M. Horschig" < jm.horschig at donders.ru.nl> wrote: > Dear Davide, > > I never worked with trials of different length, however this should be > straight forward when writing your own trialfun (just as Michael proposed). > In most trialfuns, the trl matrix is constructed by concatenating the > vectors begsample, endsample and offset. These vectors can have any values > you want them to have. As a simple example, you can write a trialfun that > just returns trl = [1 2 1; 400 500 100] (for your experiment, you of course > want these values to be computed from your trigger values). > Is your problem that such a thing does not work? It would be great if you > give a concrete use-case, or e.g. attach your trialfun. > > Best, > Jörn > > > On 8/4/2011 4:03 PM, Davide Rivolta wrote: > > Thanks Michael, > > I wrote my own trialfun. > > The offset would be negative and would have the same lenght of difference > between endsample-begsample. This, for each trial, would have a different > value. > > It can be a bug. In fact, I tried to vhange the trialfun by leaving the > same begsample and endsample, but I put a fixed offset (i.e., -100) and it > worked. All the trial was in and not, like with the variable offset, until > time = 0. > > Thanks, > > Davide > > On Thu, Aug 4, 2011 at 3:55 PM, Michael Wibral wrote: > >> Hi davide, >> >> I guess in such a case you would have to write your own trial function. >> However, offset values larger than 0 mean, that the stimulus occured before >> you start your piece of data - is that really what you wanted ? >> >> Michael >> >> >> ------------------------------ >> *Von:* "Davide Rivolta" >> *Gesendet:* Aug 4, 2011 2:44:08 PM >> *An:* fieldtrip at donders.ru.nl >> *Betreff:* [FieldTrip] Potential bug in trl >> >> >> Dear all, >> >> I have noticed that the trial definition works well when begsample, >> ensample and offset have fixed values. >> >> I have however a specific trialfunction, where each trial has a different >> lenght. >> >> This works well when you specify a fixed offset, however, when even the >> offset has to change for each trial, Fieldtrip reports values until 0. It >> ignores time after 0. >> >> >> Might it be a bug? >> >> Thanks, >> >> Davide >> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> > > > > -- > Davide Rivolta, PhD > > > > _______________________________________________ > fieldtrip mailing listfieldtrip at donders.ru.nlhttp://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > -- > Jörn M. Horschig > PhD Student > Donders Institute for Brain, Cognition and Behaviour > Centre for Cognitive Neuroimaging > Radboud University Nijmegen > Neuronal Oscillations Group > > P.O. Box 9101 > NL-6500 HB Nijmegen > The Netherlands > > Contact: > E-Mail: jm.horschig at donders.ru.nl > Tel: +31-(0)24-36-68493 > Web: http://www.ru.nl/donders > > Visiting address: > Trigon, room 2.30 > Kapittelweg 29 > NL-6525 EN Nijmegen > The Netherlands > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > -- Davide Rivolta, PhD -------------- next part -------------- An HTML attachment was scrubbed... URL: From BelluscB at ninds.nih.gov Fri Aug 5 16:20:11 2011 From: BelluscB at ninds.nih.gov (Belluscio, Beth (NIH/NINDS) [E]) Date: Fri, 5 Aug 2011 10:20:11 -0400 Subject: [FieldTrip] time frequency analysis - basic question In-Reply-To: <4E3B9BE2.4000300@donders.ru.nl> References: <794DFDD3C128EF44AC49ADC58FD0090C057456B617@NIHMLBX10.nih.gov> <4E3B9BE2.4000300@donders.ru.nl> Message-ID: <794DFDD3C128EF44AC49ADC58FD0090C057456B626@NIHMLBX10.nih.gov> Thanks, Jorn, that makes sense to me. From: "Jörn M. Horschig" [mailto:jm.horschig at donders.ru.nl] Sent: Friday, August 05, 2011 3:30 AM To: Email discussion list for the FieldTrip project Subject: Re: [FieldTrip] time frequency analysis - basic question Dear Beth, >From all my knowledge I gathered so far, I can try to answer your question (and hope that people from this list correct me if I am wrong): As often, the truth lies somewhere in between. The frequency smoothing is determined by the type and length of the taper you apply - this is similar to wavelet analysis. If you use multitapers, you can specify cfg.tapsmofrq, which determines the frequency band you want to smooth with. The FT freqanalysis method uses multiplication in the time domain instead of convolution in the frequency domain (which is equivalent). Any taper will always have a specific frequency spectrum itself, and thus it will not give you the value of a single frequency (though the output looks as if). So if you specify foi as [5 10 15], then the power at 5Hz will not be from 2.5 to 7.5Hz, but neither will it be at exactly and only 5Hz (btw, the exact center frequency might not be 5Hz, this depends on the (Rayleigh) frequency resolution of your signal). I do not know any more specifics concerning this, maybe the help of ft_freqanalysis can tell you some additonal things: http://fieldtrip.fcdonders.nl/reference/ft_freqanalysis Best, Jörn Btw, if this is not correct, else from the mailinglist should correct me asap ;) On 8/4/2011 3:23 PM, Belluscio, Beth (NIH/NINDS) [E] wrote: Hi all- When calling ft_freqanalysis you specify the foi limits and intervals - eg 5:5:30. The resulting data set appears to generate the power at discreet frequencies (5, 10, 15, 20, 25, 30). Does this actually reflect an estimate of the power over a range of frequencies (2.5-7.5, 7.5-12.5, 12.5-17.5,....) or the power at discreet frequencies with those in between not evaluated? Thanks for your help, Beth. Beth Belluscio MD-PhD Clinical Fellow Human Motor Control Section NINDS, NIH 301-402-3495 _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip -- Jörn M. Horschig PhD Student Donders Institute for Brain, Cognition and Behaviour Centre for Cognitive Neuroimaging Radboud University Nijmegen Neuronal Oscillations Group P.O. Box 9101 NL-6500 HB Nijmegen The Netherlands Contact: E-Mail: jm.horschig at donders.ru.nl Tel: +31-(0)24-36-68493 Web: http://www.ru.nl/donders Visiting address: Trigon, room 2.30 Kapittelweg 29 NL-6525 EN Nijmegen The Netherlands -------------- next part -------------- An HTML attachment was scrubbed... URL: From cas243 at georgetown.edu Sat Aug 6 17:21:47 2011 From: cas243 at georgetown.edu (Clara A. Scholl) Date: Sat, 6 Aug 2011 11:21:47 -0400 Subject: [FieldTrip] Extracting a timecourse from cluster of channels In-Reply-To: <4e34099a.c188e50a.3fc3.ffff8898SMTPIN_ADDED@mx.google.com> References: <4e34099a.c188e50a.3fc3.ffff8898SMTPIN_ADDED@mx.google.com> Message-ID: Thanks Eric! Do you think it's reasonable to take the intersection, or union, of channels over time in a given space-time cluster? Do you know of any publications that have dealt with this issue (choosing channels from a space-time cluster for timecourse extraction) that I could look at & reference? Thanks, Clara On Sat, Jul 30, 2011 at 9:31 AM, Eric Maris wrote: > Dear Clara, > >> I am using fieldtrip's cluster-based permutation test to identify a >> significant channel-time cluster showing an effect.  Now I would like >> to extract a timecourse averaged over these channels (for the purpose >> of comparing different conditions, not used to generate the cluster), >> but I'm not sure if it makes sense to do so because the channels >> belonging to the cluster change over time -- would the timecourse have >> contributions from different channels at different time points?  Do I >> need to limit my cluster to fixed channels across time? (And if so, >> how do I do that?) > > Yes, you should fix channels across time. The most important point in > selecting the channels is finding evidence for the fact that the channels > reflects a single source only. This is not a statistical issue, and > therefore there is not a p-value that can guide you in this choice. If you > have collected MEG data, the best evidence is a dipolar topography of your > effect. > > Best, > > Eric Maris > > >> >> Thanks! >> Clara >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From jan.schoffelen at donders.ru.nl Sat Aug 6 21:08:02 2011 From: jan.schoffelen at donders.ru.nl (jan-mathijs schoffelen) Date: Sat, 6 Aug 2011 21:08:02 +0200 Subject: [FieldTrip] ft_volumesegment question In-Reply-To: <621FD28E-EB95-4966-9612-B64308F79074@ucdenver.edu> References: <621FD28E-EB95-4966-9612-B64308F79074@ucdenver.edu> Message-ID: Dear Don, You're absolutely right. The correct tranfsormation matrix written to the file should be the mri.transform, not the original.transform. I'll fix this as soon as possible. Thanks for the debugging. Best wishes, Jan-Mathijs On Aug 3, 2011, at 5:47 PM, Rojas, Don wrote: > Dear Fieldtrippers, > > I use FieldTrip and SPM8 together quite frequently. I noticed recently that the segmentation output (gray, white and csf tpm) of ft_volumesegment (c1, c2, c3) does not align properly to the input mri scan. This is not the case when doing the same segmentation in SPM8. I'm using the 20110725 version of Fieldtrip and have mad certain that it and not the one in spm8 is used. My cfg and command is as follows: > > cfg.output = {'brain' 'skull' 'scalp','tpm'}; > cfg.spmversion = 'spm8'; > cfg.name = 'test'; > cfg.write = 'yes'; > cfg.coordsys = 'spm'; > seg = ft_volumesegment(cfg,mri); > > Looking at the code for ft_volumesegment, it appears that a permutation of the mri dimensions (swapping 2 and 3) is done on line 279: > > [mri,permutevec,flipflags] = align_ijk2xyz(mri); > > The mri.transform field is changed accordingly. However, the original.transform is put into the headers for the c1, c2 and c3 output (line 399). I think that this is not correct because the segmentation output has the permuted dimensions, not the dimensions of the original mri. So, the correct transform should be the mri.transform after align_ijk2xyz I think. Or maybe this is intentional behavior and I'm missing something. It can easily be checked, however, using Check Reg, in SPM8, that the output tpm do not align correctly with the original mri. If I manually write the mri.transform (from align_ijk2xyz) to the tpm output, however, they do align correctly. > > Best, > > Don > > ----------------------- > Don Rojas, Ph.D. > Associate Professor of Psychiatry > U. of Colorado Denver Anschutz Medical Campus > Director, UCD Magnetoencephalography Lab > 13001 E. 17th Pl F546 > Aurora, CO 80045 > 303-724-4994 > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip Dr. J.M. (Jan-Mathijs) Schoffelen Donders Institute for Brain, Cognition and Behaviour, Centre for Cognitive Neuroimaging, Radboud University Nijmegen, The Netherlands J.Schoffelen at donders.ru.nl Telephone: +31-24-3614793 -------------- next part -------------- An HTML attachment was scrubbed... URL: From michael.wibral at web.de Mon Aug 8 11:22:18 2011 From: michael.wibral at web.de (Michael Wibral) Date: Mon, 8 Aug 2011 11:22:18 +0200 (CEST) Subject: [FieldTrip] Potential bug in trl Message-ID: <434769618.3170746.1312795338080.JavaMail.fmail@mwmweb077> An HTML attachment was scrubbed... URL: From e.maris at donders.ru.nl Mon Aug 8 11:23:39 2011 From: e.maris at donders.ru.nl (Eric Maris) Date: Mon, 8 Aug 2011 11:23:39 +0200 Subject: [FieldTrip] Extracting a timecourse from cluster of channels In-Reply-To: References: <4e34099a.c188e50a.3fc3.ffff8898SMTPIN_ADDED@mx.google.com> Message-ID: <071401cc55ac$dea9af20$9bfd0d60$@maris@donders.ru.nl> Hi Clara, > Do you think it's reasonable to take the intersection, or union, of > channels over time in a given space-time cluster? > Do you know of any publications that have dealt with this issue > (choosing channels from a space-time cluster for timecourse > extraction) that I could look at & reference? I wish I could, but I cannot help you here. Your question belongs to the category "important but not answerable on a principled basis (at least for me)". Best, Eric Maris > Thanks, Clara > > On Sat, Jul 30, 2011 at 9:31 AM, Eric Maris wrote: > > Dear Clara, > > > >> I am using fieldtrip's cluster-based permutation test to identify a > >> significant channel-time cluster showing an effect.  Now I would like > >> to extract a timecourse averaged over these channels (for the purpose > >> of comparing different conditions, not used to generate the cluster), > >> but I'm not sure if it makes sense to do so because the channels > >> belonging to the cluster change over time -- would the timecourse have > >> contributions from different channels at different time points?  Do I > >> need to limit my cluster to fixed channels across time? (And if so, > >> how do I do that?) > > > > Yes, you should fix channels across time. The most important point in > > selecting the channels is finding evidence for the fact that the channels > > reflects a single source only. This is not a statistical issue, and > > therefore there is not a p-value that can guide you in this choice. If you > > have collected MEG data, the best evidence is a dipolar topography of your > > effect. > > > > Best, > > > > Eric Maris > > > > > >> > >> Thanks! > >> Clara > >> _______________________________________________ > >> fieldtrip mailing list > >> fieldtrip at donders.ru.nl > >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > _______________________________________________ > > fieldtrip mailing list > > fieldtrip at donders.ru.nl > > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From drivolta81 at gmail.com Mon Aug 8 12:17:45 2011 From: drivolta81 at gmail.com (Davide Rivolta) Date: Mon, 8 Aug 2011 12:17:45 +0200 Subject: [FieldTrip] Potential bug in trl In-Reply-To: <434769618.3170746.1312795338080.JavaMail.fmail@mwmweb077> References: <434769618.3170746.1312795338080.JavaMail.fmail@mwmweb077> Message-ID: Dear Michael, After your initial reply I realised that I made a (silly) mistake in the trialfun. It was not a bug indeed. My apologies for this, Davide On Mon, Aug 8, 2011 at 11:22 AM, Michael Wibral wrote: > Hi Davide, > > would you mind sharing your solution (that could potentially help others as > well)? > > Thanks, > Michael > > > > ------------------------------ > *Von:* "Davide Rivolta" > *Gesendet:* Aug 5, 2011 11:36:07 AM > *An:* "Email discussion list for the FieldTrip project" < > fieldtrip at donders.ru.nl> > *Betreff:* Re: [FieldTrip] Potential bug in trl > > > Dear Jörn, > > Thank you for your reply. I guess I have now figured it out. > > > Thanks, > Davide > > On Fri, Aug 5, 2011 at 9:13 AM, "Jörn M. Horschig" < > jm.horschig at donders.ru.nl> wrote: > >> Dear Davide, >> >> I never worked with trials of different length, however this should be >> straight forward when writing your own trialfun (just as Michael proposed). >> In most trialfuns, the trl matrix is constructed by concatenating the >> vectors begsample, endsample and offset. These vectors can have any values >> you want them to have. As a simple example, you can write a trialfun that >> just returns trl = [1 2 1; 400 500 100] (for your experiment, you of course >> want these values to be computed from your trigger values). >> Is your problem that such a thing does not work? It would be great if you >> give a concrete use-case, or e.g. attach your trialfun. >> >> Best, >> Jörn >> >> >> >> On 8/4/2011 4:03 PM, Davide Rivolta wrote: >> >> Thanks Michael, >> >> I wrote my own trialfun. >> >> The offset would be negative and would have the same lenght of difference >> between endsample-begsample. This, for each trial, would have a different >> value. >> >> It can be a bug. In fact, I tried to vhange the trialfun by leaving the >> same begsample and endsample, but I put a fixed offset (i.e., -100) and it >> worked. All the trial was in and not, like with the variable offset, until >> time = 0. >> >> Thanks, >> >> Davide >> >> On Thu, Aug 4, 2011 at 3:55 PM, Michael Wibral wrote: >> >>> Hi davide, >>> >>> I guess in such a case you would have to write your own trial function. >>> However, offset values larger than 0 mean, that the stimulus occured before >>> you start your piece of data - is that really what you wanted ? >>> >>> Michael >>> >>> >>> ------------------------------ >>> *Von:* "Davide Rivolta" >>> *Gesendet:* Aug 4, 2011 2:44:08 PM >>> *An:* fieldtrip at donders.ru.nl >>> *Betreff:* [FieldTrip] Potential bug in trl >>> >>> >>> Dear all, >>> >>> I have noticed that the trial definition works well when begsample, >>> ensample and offset have fixed values. >>> >>> I have however a specific trialfunction, where each trial has a different >>> lenght. >>> >>> This works well when you specify a fixed offset, however, when even the >>> offset has to change for each trial, Fieldtrip reports values until 0. It >>> ignores time after 0. >>> >>> >>> Might it be a bug? >>> >>> Thanks, >>> >>> Davide >>> >>> >>> _______________________________________________ >>> fieldtrip mailing list >>> fieldtrip at donders.ru.nl >>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> >> >> >> >> -- >> Davide Rivolta, PhD >> >> >> >> _______________________________________________ >> fieldtrip mailing listfieldtrip at donders.ru.nlhttp://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> >> >> >> >> -- >> Jörn M. Horschig >> PhD Student >> Donders Institute for Brain, Cognition and Behaviour >> Centre for Cognitive Neuroimaging >> Radboud University Nijmegen >> Neuronal Oscillations Group >> >> P.O. Box 9101 >> NL-6500 HB Nijmegen >> The Netherlands >> >> Contact: >> E-Mail: jm.horschig at donders.ru.nl >> Tel: +31-(0)24-36-68493 >> Web: http://www.ru.nl/donders >> >> Visiting address: >> Trigon, room 2.30 >> Kapittelweg 29 >> NL-6525 EN Nijmegen >> The Netherlands >> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > > -- > Davide Rivolta, PhD > > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > -- Davide Rivolta, PhD -------------- next part -------------- An HTML attachment was scrubbed... URL: From marco.dahmane at gmail.com Mon Aug 8 15:13:24 2011 From: marco.dahmane at gmail.com (Marco Dahmane) Date: Mon, 8 Aug 2011 14:13:24 +0100 Subject: [FieldTrip] databrowser and merged data Message-ID: Hi all, I have a rather straightforward question. I would like to be able to visualize the output of ft_timelockanalysis in butterfly mode using the databrowser function. However, since my data was made by merging several raw datasets together, the sampleinfo field was erased during the process, and databrowser doesn't like that (even though the timelock.time and timelock.avg fields are fine) What I would like to know is, since I have quite a lot of small datasets to be merged together, is there a way to make FT automatically reconstruct the sampleinfo field after merging two datasets together ? In other words, do I *have* to manually add the sampleinfo field at the end (and how come it is lost upon merging two datasets?)? And if yes, what is the easiest way to go about it? Many thanks, --Marco -------------- next part -------------- An HTML attachment was scrubbed... URL: From eelke.spaak at donders.ru.nl Mon Aug 8 15:36:11 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Mon, 8 Aug 2011 15:36:11 +0200 Subject: [FieldTrip] databrowser and merged data In-Reply-To: References: Message-ID: Dear Marco, Let me make three points in response to your question. (1) Actually, in general, timelock data structures should never contain sampleinfo, since sampleinfo describes which samples in the original recording your trials correspond to (I'm not sure about timelocked data with cfg.keeptrials='yes'). (2) However, I don't think that point (1) is relevant to your problem. More relevant is this: why do you want to use the databrowser to visualise timelocked data? The databrowser is meant to conveniently page through data with a big time axis. Usually, your timelocked data will not have a big time axis, but a small one. For this, we have the ft_singleplotER-function, or simply matlab's plot(). (3) When you provide data that lacks a sampleinfo field to a fieldtrip function that requires a sampleinfo field, the sampleinfo should be automatically constructed. (This is taken care of by ft_checkdata, which is also responsible for converting your timelocked data into raw data in case you pass it to the databrowser.) So, even if you decide that you really want to use the databrowser to display timelocked data, it should actually just work :) Are you using the latest version of fieldtrip? If so, could you provide me with some more details on the error and its circumstances? Best, Eelke 2011/8/8 Marco Dahmane : > Hi all, > I have a rather straightforward question. I would like to be able to > visualize the output of ft_timelockanalysis in butterfly mode using the > databrowser function. > However, since my data was made by merging several raw datasets together, > the sampleinfo field was erased during the process, and databrowser doesn't > like that (even though the timelock.time and timelock.avg fields are fine) > What I would like to know is, since I have quite a lot of small datasets to > be merged together, is there a way to make FT automatically reconstruct the > sampleinfo field after merging two datasets together ? In other words, do I > *have* to manually add the sampleinfo field at the end (and how come it is > lost upon merging two datasets?)? > And if yes, what is the easiest way to go about it? > Many thanks, > --Marco > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From marco.dahmane at gmail.com Mon Aug 8 16:08:11 2011 From: marco.dahmane at gmail.com (Marco Dahmane) Date: Mon, 8 Aug 2011 15:08:11 +0100 Subject: [FieldTrip] databrowser and merged data Message-ID: Hi Eelke, So in order :) 1) Fair enough. And I'm not using cfg.keeptrials='yes' anyway 2) I'm actually using databrowser because I find it really convenient to just remove and add channels, or see to which channel each line corresponds. I know I could do this using Matlab's plot, but I was just wondering why it wouldn't work with databrowser. 3) Ok this is weird then. Something that will probably help you, whenever I use databrowser on my timelocked data, I get this : Warning: the trial definition in the configuration is inconsistent with the actual data > In utilities/private/warning_once at 75 In utilities/private/fixsampleinfo at 51 In ft_checkdata at 579 In ft_databrowser at 155 Warning: failed to create sampleinfo field > In utilities/private/warning_once at 75 In utilities/private/fixsampleinfo at 83 In ft_checkdata at 579 In ft_databrowser at 155 So there must be something wrong right? And yes I'm using the latest FT... Many thanks, --Marco -------------- next part -------------- An HTML attachment was scrubbed... URL: From BelluscB at ninds.nih.gov Mon Aug 8 18:08:16 2011 From: BelluscB at ninds.nih.gov (Belluscio, Beth (NIH/NINDS) [E]) Date: Mon, 8 Aug 2011 12:08:16 -0400 Subject: [FieldTrip] power line noise reduction Message-ID: <794DFDD3C128EF44AC49ADC58FD0090C057456B635@NIHMLBX10.nih.gov> Hi all- I am evaluating power using ft_freqanalysis for resting state data. In the preprocessing step, I indicate to remove power line noise at 60 Hz as follows: cfg = []; cfg.dataset = 'resting.ds'; cfg.channel = {'MEG'}; cfg.continuous = 'yes'; cfg.detrend = 'yes'; cfg.dftfilter = 'yes'; cfg.dftfreq = [60 120]; dataCB24rest = ft_preprocessing(cfg) Then I evaluate the power at frequencies from 5 to 100 Hz as follows: cfg = []; cfg.method = 'mtmfft'; cfg.taper = 'hanning'; cfg.foi = 5:5:100; cfg.tapsmofrq = 4; cfg.output = 'pow'; freqCB24rest = ft_freqanalysis(cfg, dataCB24rest) When I visualize the result, I get a VERY large value for the power at 60 Hz in every channel. For example, see attached. What am I doing wrong? Thanks for your help, Beth. Beth Belluscio MD-PhD Clinical Fellow Human Motor Control Section NINDS, NIH 301-402-3495 -------------- next part -------------- A non-text attachment was scrubbed... Name: example.tif Type: image/tiff Size: 18858 bytes Desc: example.tif URL: From jan.schoffelen at donders.ru.nl Mon Aug 8 19:43:40 2011 From: jan.schoffelen at donders.ru.nl (jan-mathijs schoffelen) Date: Mon, 8 Aug 2011 19:43:40 +0200 Subject: [FieldTrip] power line noise reduction In-Reply-To: <794DFDD3C128EF44AC49ADC58FD0090C057456B635@NIHMLBX10.nih.gov> References: <794DFDD3C128EF44AC49ADC58FD0090C057456B635@NIHMLBX10.nih.gov> Message-ID: <0E61E958-F55D-4D1C-B3CB-7F25C9E3A63E@donders.ru.nl> Hi Beth, It seems from your approach, that you use a long stretch of data on which you apply the dftfilter. The dftfilter only works well, if the power line fluctuations are stationary in amplitude. Another way of stating it, is that the line noise really has to be a very narrowband line in the spectrum (relative to the epoch length). Typically, the dftfilter works well only up to an epoch length of ~ 10 seconds (sometimes even less if there are sources of non-stationary power line interference). In your case I would either chop up the data in shorter segments, or use a notch filter (or a lowpassfilter; I would rather go for the low frequency stuff to begin with in the resting state). Best, Jan-Mathijs On Aug 8, 2011, at 6:08 PM, Belluscio, Beth (NIH/NINDS) [E] wrote: > Hi all- > I am evaluating power using ft_freqanalysis for resting state data. > In the preprocessing step, I indicate to remove power line noise at 60 Hz as follows: > > cfg = []; > cfg.dataset = 'resting.ds'; > cfg.channel = {'MEG'}; > cfg.continuous = 'yes'; > cfg.detrend = 'yes'; > cfg.dftfilter = 'yes'; > cfg.dftfreq = [60 120]; > dataCB24rest = ft_preprocessing(cfg) > > Then I evaluate the power at frequencies from 5 to 100 Hz as follows: > > cfg = []; > cfg.method = 'mtmfft'; > cfg.taper = 'hanning'; > cfg.foi = 5:5:100; > cfg.tapsmofrq = 4; > cfg.output = 'pow'; > freqCB24rest = ft_freqanalysis(cfg, dataCB24rest) > > When I visualize the result, I get a VERY large value for the power at 60 Hz in every channel. For example, see attached. > > What am I doing wrong? > Thanks for your help, > Beth. > > > Beth Belluscio MD-PhD > Clinical Fellow > Human Motor Control Section > NINDS, NIH > 301-402-3495 > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip Dr. J.M. (Jan-Mathijs) Schoffelen Donders Institute for Brain, Cognition and Behaviour, Centre for Cognitive Neuroimaging, Radboud University Nijmegen, The Netherlands J.Schoffelen at donders.ru.nl Telephone: +31-24-3614793 From ekanal at cmu.edu Mon Aug 8 20:07:26 2011 From: ekanal at cmu.edu (Kanal Eliezer) Date: Mon, 8 Aug 2011 14:07:26 -0400 Subject: [FieldTrip] question about re-doing analysis Message-ID: <150C06A5-3B96-4F2C-B529-DB1079A36701@cmu.edu> Hello folks - Simple question here... I've been working with a dataset for a while, and it turns out that the [1 50] bandpass filter I applied earlier is removing some data that we're going to want to examine. My question is, what's the best way to go about this? For most of the data, the preprocessing consisted of discarding trials (a lot of noisy trials, unfortunately) and ICA. I've saved the data from the preprocessing stages, and my current thought is to write a script to load up each dataset, extract the "sampleinfo" field, pull that from the original raw file, apply any new filters, and then work with that. This would mitigate my needing to go through all the data and do the trial reject again (which would take weeks of work). Is there a better way to do this? Thanks - Eliezer Kanal From jan.schoffelen at donders.ru.nl Mon Aug 8 20:33:48 2011 From: jan.schoffelen at donders.ru.nl (jan-mathijs schoffelen) Date: Mon, 8 Aug 2011 20:33:48 +0200 Subject: [FieldTrip] question about re-doing analysis In-Reply-To: <150C06A5-3B96-4F2C-B529-DB1079A36701@cmu.edu> References: <150C06A5-3B96-4F2C-B529-DB1079A36701@cmu.edu> Message-ID: <80BC061E-C9FD-46B5-B3D7-289260D35533@donders.ru.nl> Dear Elli, > to load up each dataset, extract the "sampleinfo" field, pull that from the original raw file, apply any new filters, and then work with that. This would mitigate my needing to go through all the data and do the trial reject again (which would take weeks of work). Is there a better way to do this? Isn't this good enough, then ;-) ? JM Dr. J.M. (Jan-Mathijs) Schoffelen Donders Institute for Brain, Cognition and Behaviour, Centre for Cognitive Neuroimaging, Radboud University Nijmegen, The Netherlands J.Schoffelen at donders.ru.nl Telephone: +31-24-3614793 From ekanal at cmu.edu Mon Aug 8 20:58:08 2011 From: ekanal at cmu.edu (Kanal Eliezer) Date: Mon, 8 Aug 2011 14:58:08 -0400 Subject: [FieldTrip] question about re-doing analysis In-Reply-To: <80BC061E-C9FD-46B5-B3D7-289260D35533@donders.ru.nl> References: <150C06A5-3B96-4F2C-B529-DB1079A36701@cmu.edu> <80BC061E-C9FD-46B5-B3D7-289260D35533@donders.ru.nl> Message-ID: <024FF5CF-ECE6-44ED-9C35-2D7636D797CF@cmu.edu> Yes, that works... the only problem with this approach is that I'll have to redo ICA filtering. I guess I was just curious if there's a different way that everyone uses that I'm not familiar with. This is the first dataset I've analyzed using fieldtrip, so I'm just trying to make sure I'm "doing it right". Thanks - Elli On Aug 8, 2011, at 2:33 PM, jan-mathijs schoffelen wrote: > Dear Elli, > >> to load up each dataset, extract the "sampleinfo" field, pull that from the original raw file, apply any new filters, and then work with that. This would mitigate my needing to go through all the data and do the trial reject again (which would take weeks of work). Is there a better way to do this? > > Isn't this good enough, then ;-) ? > > JM > > > > Dr. J.M. (Jan-Mathijs) Schoffelen > Donders Institute for Brain, Cognition and Behaviour, > Centre for Cognitive Neuroimaging, > Radboud University Nijmegen, The Netherlands > J.Schoffelen at donders.ru.nl > Telephone: +31-24-3614793 > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From jan.schoffelen at donders.ru.nl Mon Aug 8 21:05:12 2011 From: jan.schoffelen at donders.ru.nl (jan-mathijs schoffelen) Date: Mon, 8 Aug 2011 21:05:12 +0200 Subject: [FieldTrip] question about re-doing analysis In-Reply-To: <024FF5CF-ECE6-44ED-9C35-2D7636D797CF@cmu.edu> References: <150C06A5-3B96-4F2C-B529-DB1079A36701@cmu.edu> <80BC061E-C9FD-46B5-B3D7-289260D35533@donders.ru.nl> <024FF5CF-ECE6-44ED-9C35-2D7636D797CF@cmu.edu> Message-ID: <6044F712-0B26-4637-B05B-9E89E9A7D206@donders.ru.nl> Hi Elli, Yes, that's a good point. What I sometimes do, is to also store the mixing matrix which comes out after ft_componentanalysis (i.e. the comp.topo). When you have this matrix, you can very cheaply re-ICA-filter the data. Yet, note that with different filter settings in the preprocessing, the ICA-components (and topographies) may also change. BW, JM On Aug 8, 2011, at 8:58 PM, Kanal Eliezer wrote: > Yes, that works... the only problem with this approach is that I'll have to redo ICA filtering. > > I guess I was just curious if there's a different way that everyone uses that I'm not familiar with. This is the first dataset I've analyzed using fieldtrip, so I'm just trying to make sure I'm "doing it right". Thanks - > > Elli > > > On Aug 8, 2011, at 2:33 PM, jan-mathijs schoffelen wrote: > >> Dear Elli, >> >>> to load up each dataset, extract the "sampleinfo" field, pull that from the original raw file, apply any new filters, and then work with that. This would mitigate my needing to go through all the data and do the trial reject again (which would take weeks of work). Is there a better way to do this? >> >> Isn't this good enough, then ;-) ? >> >> JM >> >> >> >> Dr. J.M. (Jan-Mathijs) Schoffelen >> Donders Institute for Brain, Cognition and Behaviour, >> Centre for Cognitive Neuroimaging, >> Radboud University Nijmegen, The Netherlands >> J.Schoffelen at donders.ru.nl >> Telephone: +31-24-3614793 >> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip Dr. J.M. (Jan-Mathijs) Schoffelen Donders Institute for Brain, Cognition and Behaviour, Centre for Cognitive Neuroimaging, Radboud University Nijmegen, The Netherlands J.Schoffelen at donders.ru.nl Telephone: +31-24-3614793 From ekanal at cmu.edu Mon Aug 8 21:19:42 2011 From: ekanal at cmu.edu (Kanal Eliezer) Date: Mon, 8 Aug 2011 15:19:42 -0400 Subject: [FieldTrip] question about re-doing analysis In-Reply-To: <6044F712-0B26-4637-B05B-9E89E9A7D206@donders.ru.nl> References: <150C06A5-3B96-4F2C-B529-DB1079A36701@cmu.edu> <80BC061E-C9FD-46B5-B3D7-289260D35533@donders.ru.nl> <024FF5CF-ECE6-44ED-9C35-2D7636D797CF@cmu.edu> <6044F712-0B26-4637-B05B-9E89E9A7D206@donders.ru.nl> Message-ID: <5B688DA3-CECE-42F5-9E0C-70F3FCB086FB@cmu.edu> That's a good point, I guess I'll have needed to re-do them anyway. Elli On Aug 8, 2011, at 3:05 PM, jan-mathijs schoffelen wrote: > Hi Elli, > > Yes, that's a good point. What I sometimes do, is to also store the mixing matrix which comes out after ft_componentanalysis (i.e. the comp.topo). When you have this matrix, you can very cheaply re-ICA-filter the data. Yet, note that with different filter settings in the preprocessing, the ICA-components (and topographies) may also change. > > > BW, JM > > > > On Aug 8, 2011, at 8:58 PM, Kanal Eliezer wrote: > >> Yes, that works... the only problem with this approach is that I'll have to redo ICA filtering. >> >> I guess I was just curious if there's a different way that everyone uses that I'm not familiar with. This is the first dataset I've analyzed using fieldtrip, so I'm just trying to make sure I'm "doing it right". Thanks - >> >> Elli >> >> >> On Aug 8, 2011, at 2:33 PM, jan-mathijs schoffelen wrote: >> >>> Dear Elli, >>> >>>> to load up each dataset, extract the "sampleinfo" field, pull that from the original raw file, apply any new filters, and then work with that. This would mitigate my needing to go through all the data and do the trial reject again (which would take weeks of work). Is there a better way to do this? >>> >>> Isn't this good enough, then ;-) ? >>> >>> JM >>> >>> >>> >>> Dr. J.M. (Jan-Mathijs) Schoffelen >>> Donders Institute for Brain, Cognition and Behaviour, >>> Centre for Cognitive Neuroimaging, >>> Radboud University Nijmegen, The Netherlands >>> J.Schoffelen at donders.ru.nl >>> Telephone: +31-24-3614793 >>> >>> >>> _______________________________________________ >>> fieldtrip mailing list >>> fieldtrip at donders.ru.nl >>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > Dr. J.M. (Jan-Mathijs) Schoffelen > Donders Institute for Brain, Cognition and Behaviour, > Centre for Cognitive Neuroimaging, > Radboud University Nijmegen, The Netherlands > J.Schoffelen at donders.ru.nl > Telephone: +31-24-3614793 > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From eelke.spaak at donders.ru.nl Tue Aug 9 11:00:27 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Tue, 9 Aug 2011 11:00:27 +0200 Subject: [FieldTrip] databrowser and merged data In-Reply-To: References: Message-ID: Hi Marco, The subfunction fixsampleinfo (used by ft_checkdata) attempts a number of things in order to create a sampleinfo field. My guess is that it detected a trial definition somewhere in your cfg "tree" (the data.cfg and data.cfg.previous fields) that was inconsistent with your data, and therefore refused to continue rebuilding a sampleinfo from scratch. I have now changed this function so that it will always reconstruct your sampleinfo, even if it detects an inconsistent trial definition in a previous configuration structure. So, it should now work :) At least, with tomorrow's update on the FTP server. Best, Eelke 2011/8/8 Marco Dahmane : > Hi Eelke, > So in order :) > 1) Fair enough. And I'm not using cfg.keeptrials='yes' anyway > 2) I'm actually using databrowser because I find it really convenient to > just remove and add channels, or see to which channel each line corresponds. > I know I could do this using Matlab's plot, but I was just wondering why it > wouldn't work with databrowser. > 3) Ok this is weird then. Something that will probably help you, whenever I > use databrowser on my timelocked data, I get this : > Warning: the trial definition in the configuration is inconsistent with the > actual data >> In utilities/private/warning_once at 75 >   In utilities/private/fixsampleinfo at 51 >   In ft_checkdata at 579 >   In ft_databrowser at 155 > Warning: failed to create sampleinfo field >> In utilities/private/warning_once at 75 >   In utilities/private/fixsampleinfo at 83 >   In ft_checkdata at 579 >   In ft_databrowser at 155 >  So there must be something wrong right? > And yes I'm using the latest FT... > Many thanks, > --Marco > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From jm.horschig at donders.ru.nl Tue Aug 9 12:21:45 2011 From: jm.horschig at donders.ru.nl (=?ISO-8859-1?Q?=22J=F6rn_M=2E_Horschig=22?=) Date: Tue, 09 Aug 2011 12:21:45 +0200 Subject: [FieldTrip] databrowser and merged data In-Reply-To: References: Message-ID: <4E410A39.2060809@donders.ru.nl> Hi Marco and Eelke, sampleinfo cannot be reconstructed after resampling the data, because there is no way to precisely relate your resampled samples to the original samples. My guess would be that you, Marco, did that and therefore encounter the problem you mentioned. If that was so, I am not sure whether fixsampleinfo should actually recompute the sampleinfo field, because it makes absolutely no sense to have a sampleinfo field in your resampled data that relates back samples in the original data. Anyway, if you did not resample your data, then fixsampleinfo should work. Could you let us know how you preprocessed your data? Best, Jörn On 8/9/2011 11:00 AM, Eelke Spaak wrote: > Hi Marco, > > The subfunction fixsampleinfo (used by ft_checkdata) attempts a number > of things in order to create a sampleinfo field. My guess is that it > detected a trial definition somewhere in your cfg "tree" (the data.cfg > and data.cfg.previous fields) that was inconsistent with your data, > and therefore refused to continue rebuilding a sampleinfo from > scratch. I have now changed this function so that it will always > reconstruct your sampleinfo, even if it detects an inconsistent trial > definition in a previous configuration structure. > > So, it should now work :) At least, with tomorrow's update on the FTP server. > > Best, > Eelke > > 2011/8/8 Marco Dahmane: >> Hi Eelke, >> So in order :) >> 1) Fair enough. And I'm not using cfg.keeptrials='yes' anyway >> 2) I'm actually using databrowser because I find it really convenient to >> just remove and add channels, or see to which channel each line corresponds. >> I know I could do this using Matlab's plot, but I was just wondering why it >> wouldn't work with databrowser. >> 3) Ok this is weird then. Something that will probably help you, whenever I >> use databrowser on my timelocked data, I get this : >> Warning: the trial definition in the configuration is inconsistent with the >> actual data >>> In utilities/private/warning_once at 75 >> In utilities/private/fixsampleinfo at 51 >> In ft_checkdata at 579 >> In ft_databrowser at 155 >> Warning: failed to create sampleinfo field >>> In utilities/private/warning_once at 75 >> In utilities/private/fixsampleinfo at 83 >> In ft_checkdata at 579 >> In ft_databrowser at 155 >> So there must be something wrong right? >> And yes I'm using the latest FT... >> Many thanks, >> --Marco >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip -- Jörn M. Horschig PhD Student Donders Institute for Brain, Cognition and Behaviour Centre for Cognitive Neuroimaging Radboud University Nijmegen Neuronal Oscillations Group P.O. Box 9101 NL-6500 HB Nijmegen The Netherlands Contact: E-Mail: jm.horschig at donders.ru.nl Tel: +31-(0)24-36-68493 Web: http://www.ru.nl/donders Visiting address: Trigon, room 2.30 Kapittelweg 29 NL-6525 EN Nijmegen The Netherlands From eelke.spaak at donders.ru.nl Tue Aug 9 12:28:21 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Tue, 9 Aug 2011 12:28:21 +0200 Subject: [FieldTrip] databrowser and merged data In-Reply-To: <4E410A39.2060809@donders.ru.nl> References: <4E410A39.2060809@donders.ru.nl> Message-ID: Hi Jörn, As you can read in my email, I actually changed fixsampleinfo so that it *does* reconstruct the sample info now :) I agree that sampleinfo does not make sense in the case of data that is (1) timelock-averaged and (2) combined from multiple datasets, but the issue here was to get the databrowser working on such data. Since the data used internally by the databrowser (after the call to ft_checkdata) is never exposed to the user, this seems like a good solution to me. Best, Eelke 2011/8/9 "Jörn M. Horschig" : > Hi Marco and Eelke, > > sampleinfo cannot be reconstructed after resampling the data, because there > is no way to precisely relate your resampled samples to the original > samples. My guess would be that you, Marco, did that and therefore encounter > the problem you mentioned. If that was so, I am not sure whether > fixsampleinfo should actually recompute the sampleinfo field, because it > makes absolutely no sense to have a sampleinfo field in your resampled data > that relates back samples in the original data. > Anyway, if you did not resample your data, then fixsampleinfo should work. > Could you let us know how you preprocessed your data? > > Best, > Jörn > > On 8/9/2011 11:00 AM, Eelke Spaak wrote: >> >> Hi Marco, >> >> The subfunction fixsampleinfo (used by ft_checkdata) attempts a number >> of things in order to create a sampleinfo field. My guess is that it >> detected a trial definition somewhere in your cfg "tree" (the data.cfg >> and data.cfg.previous fields) that was inconsistent with your data, >> and therefore refused to continue rebuilding a sampleinfo from >> scratch. I have now changed this function so that it will always >> reconstruct your sampleinfo, even if it detects an inconsistent trial >> definition in a previous configuration structure. >> >> So, it should now work :) At least, with tomorrow's update on the FTP >> server. >> >> Best, >> Eelke >> >> 2011/8/8 Marco Dahmane: >>> >>> Hi Eelke, >>> So in order :) >>> 1) Fair enough. And I'm not using cfg.keeptrials='yes' anyway >>> 2) I'm actually using databrowser because I find it really convenient to >>> just remove and add channels, or see to which channel each line >>> corresponds. >>> I know I could do this using Matlab's plot, but I was just wondering why >>> it >>> wouldn't work with databrowser. >>> 3) Ok this is weird then. Something that will probably help you, whenever >>> I >>> use databrowser on my timelocked data, I get this : >>> Warning: the trial definition in the configuration is inconsistent with >>> the >>> actual data >>>> >>>> In utilities/private/warning_once at 75 >>> >>>   In utilities/private/fixsampleinfo at 51 >>>   In ft_checkdata at 579 >>>   In ft_databrowser at 155 >>> Warning: failed to create sampleinfo field >>>> >>>> In utilities/private/warning_once at 75 >>> >>>   In utilities/private/fixsampleinfo at 83 >>>   In ft_checkdata at 579 >>>   In ft_databrowser at 155 >>>  So there must be something wrong right? >>> And yes I'm using the latest FT... >>> Many thanks, >>> --Marco >>> _______________________________________________ >>> fieldtrip mailing list >>> fieldtrip at donders.ru.nl >>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >>> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > -- > Jörn M. Horschig > PhD Student > Donders Institute for Brain, Cognition and Behaviour > Centre for Cognitive Neuroimaging > Radboud University Nijmegen > Neuronal Oscillations Group > > P.O. Box 9101 > NL-6500 HB Nijmegen > The Netherlands > > Contact: > E-Mail: jm.horschig at donders.ru.nl > Tel:    +31-(0)24-36-68493 > Web: http://www.ru.nl/donders > > Visiting address: > Trigon, room 2.30 > Kapittelweg 29 > NL-6525 EN Nijmegen > The Netherlands > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From eelke.spaak at donders.ru.nl Tue Aug 9 12:30:07 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Tue, 9 Aug 2011 12:30:07 +0200 Subject: [FieldTrip] databrowser and merged data In-Reply-To: References: <4E410A39.2060809@donders.ru.nl> Message-ID: PS: In other words, fixsampleinfo now generates an ad-hoc sampleinfo with no relation to the original dataset, if requested to do so. Marco's preprocessing steps are not relevant for this. 2011/8/9 Eelke Spaak : > Hi Jörn, > > As you can read in my email, I actually changed fixsampleinfo so that > it *does* reconstruct the sample info now :) > > I agree that sampleinfo does not make sense in the case of data that > is (1) timelock-averaged and (2) combined from multiple datasets, but > the issue here was to get the databrowser working on such data. Since > the data used internally by the databrowser (after the call to > ft_checkdata) is never exposed to the user, this seems like a good > solution to me. > > Best, > Eelke > > 2011/8/9 "Jörn M. Horschig" : >> Hi Marco and Eelke, >> >> sampleinfo cannot be reconstructed after resampling the data, because there >> is no way to precisely relate your resampled samples to the original >> samples. My guess would be that you, Marco, did that and therefore encounter >> the problem you mentioned. If that was so, I am not sure whether >> fixsampleinfo should actually recompute the sampleinfo field, because it >> makes absolutely no sense to have a sampleinfo field in your resampled data >> that relates back samples in the original data. >> Anyway, if you did not resample your data, then fixsampleinfo should work. >> Could you let us know how you preprocessed your data? >> >> Best, >> Jörn >> >> On 8/9/2011 11:00 AM, Eelke Spaak wrote: >>> >>> Hi Marco, >>> >>> The subfunction fixsampleinfo (used by ft_checkdata) attempts a number >>> of things in order to create a sampleinfo field. My guess is that it >>> detected a trial definition somewhere in your cfg "tree" (the data.cfg >>> and data.cfg.previous fields) that was inconsistent with your data, >>> and therefore refused to continue rebuilding a sampleinfo from >>> scratch. I have now changed this function so that it will always >>> reconstruct your sampleinfo, even if it detects an inconsistent trial >>> definition in a previous configuration structure. >>> >>> So, it should now work :) At least, with tomorrow's update on the FTP >>> server. >>> >>> Best, >>> Eelke >>> >>> 2011/8/8 Marco Dahmane: >>>> >>>> Hi Eelke, >>>> So in order :) >>>> 1) Fair enough. And I'm not using cfg.keeptrials='yes' anyway >>>> 2) I'm actually using databrowser because I find it really convenient to >>>> just remove and add channels, or see to which channel each line >>>> corresponds. >>>> I know I could do this using Matlab's plot, but I was just wondering why >>>> it >>>> wouldn't work with databrowser. >>>> 3) Ok this is weird then. Something that will probably help you, whenever >>>> I >>>> use databrowser on my timelocked data, I get this : >>>> Warning: the trial definition in the configuration is inconsistent with >>>> the >>>> actual data >>>>> >>>>> In utilities/private/warning_once at 75 >>>> >>>>   In utilities/private/fixsampleinfo at 51 >>>>   In ft_checkdata at 579 >>>>   In ft_databrowser at 155 >>>> Warning: failed to create sampleinfo field >>>>> >>>>> In utilities/private/warning_once at 75 >>>> >>>>   In utilities/private/fixsampleinfo at 83 >>>>   In ft_checkdata at 579 >>>>   In ft_databrowser at 155 >>>>  So there must be something wrong right? >>>> And yes I'm using the latest FT... >>>> Many thanks, >>>> --Marco >>>> _______________________________________________ >>>> fieldtrip mailing list >>>> fieldtrip at donders.ru.nl >>>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >>>> >>> _______________________________________________ >>> fieldtrip mailing list >>> fieldtrip at donders.ru.nl >>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> >> >> -- >> Jörn M. Horschig >> PhD Student >> Donders Institute for Brain, Cognition and Behaviour >> Centre for Cognitive Neuroimaging >> Radboud University Nijmegen >> Neuronal Oscillations Group >> >> P.O. Box 9101 >> NL-6500 HB Nijmegen >> The Netherlands >> >> Contact: >> E-Mail: jm.horschig at donders.ru.nl >> Tel:    +31-(0)24-36-68493 >> Web: http://www.ru.nl/donders >> >> Visiting address: >> Trigon, room 2.30 >> Kapittelweg 29 >> NL-6525 EN Nijmegen >> The Netherlands >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> > From jm.horschig at donders.ru.nl Tue Aug 9 14:11:34 2011 From: jm.horschig at donders.ru.nl (=?ISO-8859-1?Q?=22J=F6rn_M=2E_Horschig=22?=) Date: Tue, 09 Aug 2011 14:11:34 +0200 Subject: [FieldTrip] databrowser and merged data In-Reply-To: References: <4E410A39.2060809@donders.ru.nl> Message-ID: <4E4123F6.9000209@donders.ru.nl> Hi Eelke, I have read your mail and would like to emphasize that I disagree. If a sampleinfo field is constructed that cannot be used to infer anything from, it should not be done. Thus, the fix should not be in fixsampleinfo, because it will result in all kind of other issues when it comes to reconstructing sampleinfo, although it might help to solve this particular problem. Let's discuss this at the next FT meeting rather than here :) Best, Jörn On 8/9/2011 12:30 PM, Eelke Spaak wrote: > PS: In other words, fixsampleinfo now generates an ad-hoc sampleinfo > with no relation to the original dataset, if requested to do so. > Marco's preprocessing steps are not relevant for this. > > 2011/8/9 Eelke Spaak: >> Hi Jörn, >> >> As you can read in my email, I actually changed fixsampleinfo so that >> it *does* reconstruct the sample info now :) >> >> I agree that sampleinfo does not make sense in the case of data that >> is (1) timelock-averaged and (2) combined from multiple datasets, but >> the issue here was to get the databrowser working on such data. Since >> the data used internally by the databrowser (after the call to >> ft_checkdata) is never exposed to the user, this seems like a good >> solution to me. >> >> Best, >> Eelke >> >> 2011/8/9 "Jörn M. Horschig": >>> Hi Marco and Eelke, >>> >>> sampleinfo cannot be reconstructed after resampling the data, because there >>> is no way to precisely relate your resampled samples to the original >>> samples. My guess would be that you, Marco, did that and therefore encounter >>> the problem you mentioned. If that was so, I am not sure whether >>> fixsampleinfo should actually recompute the sampleinfo field, because it >>> makes absolutely no sense to have a sampleinfo field in your resampled data >>> that relates back samples in the original data. >>> Anyway, if you did not resample your data, then fixsampleinfo should work. >>> Could you let us know how you preprocessed your data? >>> >>> Best, >>> Jörn >>> >>> On 8/9/2011 11:00 AM, Eelke Spaak wrote: >>>> Hi Marco, >>>> >>>> The subfunction fixsampleinfo (used by ft_checkdata) attempts a number >>>> of things in order to create a sampleinfo field. My guess is that it >>>> detected a trial definition somewhere in your cfg "tree" (the data.cfg >>>> and data.cfg.previous fields) that was inconsistent with your data, >>>> and therefore refused to continue rebuilding a sampleinfo from >>>> scratch. I have now changed this function so that it will always >>>> reconstruct your sampleinfo, even if it detects an inconsistent trial >>>> definition in a previous configuration structure. >>>> >>>> So, it should now work :) At least, with tomorrow's update on the FTP >>>> server. >>>> >>>> Best, >>>> Eelke >>>> >>>> 2011/8/8 Marco Dahmane: >>>>> Hi Eelke, >>>>> So in order :) >>>>> 1) Fair enough. And I'm not using cfg.keeptrials='yes' anyway >>>>> 2) I'm actually using databrowser because I find it really convenient to >>>>> just remove and add channels, or see to which channel each line >>>>> corresponds. >>>>> I know I could do this using Matlab's plot, but I was just wondering why >>>>> it >>>>> wouldn't work with databrowser. >>>>> 3) Ok this is weird then. Something that will probably help you, whenever >>>>> I >>>>> use databrowser on my timelocked data, I get this : >>>>> Warning: the trial definition in the configuration is inconsistent with >>>>> the >>>>> actual data >>>>>> In utilities/private/warning_once at 75 >>>>> In utilities/private/fixsampleinfo at 51 >>>>> In ft_checkdata at 579 >>>>> In ft_databrowser at 155 >>>>> Warning: failed to create sampleinfo field >>>>>> In utilities/private/warning_once at 75 >>>>> In utilities/private/fixsampleinfo at 83 >>>>> In ft_checkdata at 579 >>>>> In ft_databrowser at 155 >>>>> So there must be something wrong right? >>>>> And yes I'm using the latest FT... >>>>> Many thanks, >>>>> --Marco >>>>> _______________________________________________ >>>>> fieldtrip mailing list >>>>> fieldtrip at donders.ru.nl >>>>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >>>>> >>>> _______________________________________________ >>>> fieldtrip mailing list >>>> fieldtrip at donders.ru.nl >>>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >>> >>> -- >>> Jörn M. Horschig >>> PhD Student >>> Donders Institute for Brain, Cognition and Behaviour >>> Centre for Cognitive Neuroimaging >>> Radboud University Nijmegen >>> Neuronal Oscillations Group >>> >>> P.O. Box 9101 >>> NL-6500 HB Nijmegen >>> The Netherlands >>> >>> Contact: >>> E-Mail: jm.horschig at donders.ru.nl >>> Tel: +31-(0)24-36-68493 >>> Web: http://www.ru.nl/donders >>> >>> Visiting address: >>> Trigon, room 2.30 >>> Kapittelweg 29 >>> NL-6525 EN Nijmegen >>> The Netherlands >>> >>> _______________________________________________ >>> fieldtrip mailing list >>> fieldtrip at donders.ru.nl >>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >>> > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip -- Jörn M. Horschig PhD Student Donders Institute for Brain, Cognition and Behaviour Centre for Cognitive Neuroimaging Radboud University Nijmegen Neuronal Oscillations Group P.O. Box 9101 NL-6500 HB Nijmegen The Netherlands Contact: E-Mail: jm.horschig at donders.ru.nl Tel: +31-(0)24-36-68493 Web: http://www.ru.nl/donders Visiting address: Trigon, room 2.30 Kapittelweg 29 NL-6525 EN Nijmegen The Netherlands From hubert.preissl at uni-tuebingen.de Wed Aug 10 11:16:08 2011 From: hubert.preissl at uni-tuebingen.de (Hubert Preissl) Date: Wed, 10 Aug 2011 11:16:08 +0200 Subject: [FieldTrip] =?windows-1252?q?Autumn_School_=93The_multimodal_brai?= =?windows-1252?q?n=94=3A_5th-6th_October_2011=2C_T=FCbingen?= In-Reply-To: <5B688DA3-CECE-42F5-9E0C-70F3FCB086FB@cmu.edu> References: <150C06A5-3B96-4F2C-B529-DB1079A36701@cmu.edu> <80BC061E-C9FD-46B5-B3D7-289260D35533@donders.ru.nl> <024FF5CF-ECE6-44ED-9C35-2D7636D797CF@cmu.edu> <6044F712-0B26-4637-B05B-9E89E9A7D206@donders.ru.nl> <5B688DA3-CECE-42F5-9E0C-70F3FCB086FB@cmu.edu> Message-ID: <4E424C58.6000406@uni-tuebingen.de> *Autumn School “The multimodal brain”: 5^th -6^th October 2011* Hello, we are pleased to announce the upcoming “Autumn school” in Tübingen organized by the MEG Center. For application and further information on invited speakers and scientific program please visit our website: http://www.mp.uni-tuebingen.de/mp/index.php?id=396 We look forward to meet you in Tübingen! Best regards , Hubert Preissl ps. sorry if you receive this mail several times based on some cross-posting. -- Dr. Hubert Preissl MEG Center phone: ++49-(0)7071-2987704 Otfried Müller Str 47 fax: ++49-(0)7071-295706 72076 Tübingen, Germany email: hubert.preissl at uni-tuebingen.de -------------- next part -------------- An HTML attachment was scrubbed... URL: From sylvana.schister at utah.edu Fri Aug 12 00:56:44 2011 From: sylvana.schister at utah.edu (Sylvana Schister) Date: Thu, 11 Aug 2011 16:56:44 -0600 Subject: [FieldTrip] Alignment of MRI and estimated sources Message-ID: <1313103404.26076.16.camel@sylvana-desktop> Hello Fieldtrippers! I am having trouble getting meaningful results after performing source analysis with DICS due to a misalignment of my MRI dataset. I am relatively new to FT and I will really appreciate your help. I am using the 'standard_BEM' files provided in the tutorials. I manage to realign the MRI to head coordinates using volumerealign, reslice and segment it. Then, I use the modified MRI to compute the forward model. The problem arises when proceeding to ft_sourceinterpolate and ft_sourceanalysis. As I understand, the 'mri' input to ft_sourceinterpolate must be the output of FT_READ_MRI or the filename of a MRI. The function crashes when using my realigned version of the data, therefore I am using the original file, 'standard_mri.mat'. This file contains no fiducials. No surprise the results are then messed up. How/ where do I need to specify the fiducials? Why can't I use the aligned mri? What is the correct way to do this? Thank you so much for your help! Sylvana From Santosh.VemaKrishnamurthy at nrc-cnrc.gc.ca Fri Aug 12 16:00:10 2011 From: Santosh.VemaKrishnamurthy at nrc-cnrc.gc.ca (Vema Krishnamurthy, Santosh) Date: Fri, 12 Aug 2011 07:00:10 -0700 Subject: [FieldTrip] Scalp segmentation from an MRI (.nii or .fif) Message-ID: Hello Everyone, I'm very new to Fieldtrip and I need to perform scalp segmentation from an mri file(.nii or .fif). I'm working on a project where I need the scalp surface for surface matching, so I need the scalp surface with high resolution which clearly shows the facial features (nose, eyes, lips, chin). Any help/suggestions relating to this are much appreciated. Thanks and Regards, --------------------------------------------------- Santosh Vema KrishnaMurthy, M.A.Sc Medical Devices Engineer NRC Institute for Biodiagnostics (Atlantic) Elekta/IWK MEG Research office Ph: (902) 470-3984 --------------------------------------------------- -------------- next part -------------- An HTML attachment was scrubbed... URL: From Gregor.Volberg at psychologie.uni-regensburg.de Fri Aug 12 18:47:30 2011 From: Gregor.Volberg at psychologie.uni-regensburg.de (Gregor Volberg) Date: Fri, 12 Aug 2011 18:47:30 +0200 Subject: [FieldTrip] Antw: Scalp segmentation from an MRI (.nii or .fif) Message-ID: <4E457543020000570000A631@gwsmtp1.uni-regensburg.de> Dear Santosh, for that purpose you might want to have a look at the NFT toolbox from SCCN, http://sccn.ucsd.edu/nft/ . It does high resolution segmentations for brain, csf, scalp and skin surfaces. If you need the segmentation for computing forward solutions with BEM head models, then I would discourage a too fine resolution, though. It strongly increases the computational load. Best regards, Gregor -- Dr. rer. nat. Gregor Volberg ( mailto:gregor.volberg at psychologie.uni-regensburg.de ) University of Regensburg Institute for Experimental Psychology 93040 Regensburg, Germany Tel: +49 941 943 3862 Fax: +49 941 943 3233 http://www.psychologie.uni-regensburg.de/Greenlee/team/volberg/volberg.html >>> "Vema Krishnamurthy, Santosh" 12.08.11 16.42 Uhr >>> Hello Everyone, I'm very new to Fieldtrip and I need to perform scalp segmentation from an mri file(.nii or .fif). I'm working on a project where I need the scalp surface for surface matching, so I need the scalp surface with high resolution which clearly shows the facial features (nose, eyes, lips, chin). Any help/suggestions relating to this are much appreciated. Thanks and Regards, --------------------------------------------------- Santosh Vema KrishnaMurthy, M.A.Sc Medical Devices Engineer NRC Institute for Biodiagnostics (Atlantic) Elekta/IWK MEG Research office Ph: (902) 470-3984 --------------------------------------------------- -------------- next part -------------- An HTML attachment was scrubbed... URL: From jan.schoffelen at donders.ru.nl Sat Aug 13 18:23:49 2011 From: jan.schoffelen at donders.ru.nl (jan-mathijs schoffelen) Date: Sat, 13 Aug 2011 18:23:49 +0200 Subject: [FieldTrip] Fwd: ft_definetrial crash References: Message-ID: <4B01B4FC-FF35-435F-8167-F2A82C0D1B9E@donders.ru.nl> Hi Tobias, I forward your question to the discussion list, so that many may benefit from the discussion. I am not sure what is going on and I don't know whether other people have encountered this. Are you calling the function from the correct directory? In other words, you should not be within the *.ds directory itself. You could also try to explicitly define cfg.datafile and cfg.headerfile (and omit the cfg.dataset). Let me know when the problem persists. Best wishes, Jan-Mathijs Begin forwarded message: > From: Tobias Overath > Date: August 3, 2011 12:37:15 PM GMT+02:00 > To: jan.schoffelen at donders.ru.nl > Subject: ft_definetrial crash > > Hi Jan-Mathijs, > > I participated in the MEG FieldTrip course this past April and have finally managed to get my first MEG dataset from the CTF machine here at the FIL. However, I am already crashing at the first step, i.e. when running ft_definetrial. > > This is my code: > cfg = []; > cfg.dataset = 'S01_01.ds'; > cfg.trialdef.eventtype = 'frontpanel trigger'; > cfg.trialdef.eventvalue = 1; > cfg.trialdef.prestim = .25; > cfg.trialdef.poststim = 2.75; > cfg = ft_definetrial(cfg); > > It recognises correctly that it's supposed to read the header from S01_01.ds/S01_01.res, but it then tells me that it can't find the file S01_01.meg4 (which is true, because it should be S01_01.ds/S01_01.meg4). I presume this is a pretty common error, but it occurs with two FieldTrip versions that I have downloaded (very recent ones). I am probably missing something very obvious... > > Incidentally, is there no other way to search the FieldTrip email archives than to download a zipped text file for each month of the previous 7 years?!? > > Thanks, > > Tobias > > > > -- > Tobias Overath, PhD > UCL Ear Institute > University College London > 332 Grays Inn Road > London, WC1X 8EE > UK > http://www.homepages.ucl.ac.uk/~skgtjto > > > > > Dr. J.M. (Jan-Mathijs) Schoffelen Donders Institute for Brain, Cognition and Behaviour, Centre for Cognitive Neuroimaging, Radboud University Nijmegen, The Netherlands J.Schoffelen at donders.ru.nl Telephone: +31-24-3614793 -------------- next part -------------- An HTML attachment was scrubbed... URL: From r.oostenveld at donders.ru.nl Mon Aug 15 10:10:06 2011 From: r.oostenveld at donders.ru.nl (Robert Oostenveld) Date: Mon, 15 Aug 2011 10:10:06 +0200 Subject: [FieldTrip] Invitation to contribute to a Special Issue on Brain Oscillations during Language Processing References: Message-ID: <021E015B-2CB1-46FA-96E3-0DA3922CEB58@donders.ru.nl> Begin forwarded message: > Dear colleagues, > We would like to invite you to contribute your research to our special issue on the role of brain oscillations in language processing, to appear in Frontiers in Language Science. > You can visit the web site at: > http://www.frontiersin.org/languagesciences/specialtopics/brain_oscillations_during_lang/397 > See detailed description below. > The call has been very successful so far and prominent figures in the field have joined us in this project. > We are looking forward to receiving your research. > Best wishes, > Lucia Melloni & Marcela Pena > > > Brain Oscillations during Language Processing: from Perception to Production > > Deadline for abstract submission: 01 Sep 2011 > Deadline for full article submission: 15 Dec 2011 > > Language processing is a seemingly effortless task that requires the integration of speech units (e.g., phonemes, syllables, words, etc.) occurring at different rates. In particular, temporal binding for speech should occur within and across different temporal scales, necessitating multiple simultaneous windows of integration for prosodic, semantic, syntactic and pragmatic processing. Recent evidence suggests that neuronal oscillations may reflect both tracking linguistic units at their individual rhythms as well as integrating speech units over a large range of temporal scales. > > The present Research Topic would like to evaluate current theories and evidence for a mechanistic role of neuronal oscillations in measuring language processing, covering the latest advances brought about by EEG, MEG and fMRI imaging methods. Our main focus is to highlight innovative and foundational studies that go beyond methodological issues and advance our theoretical understanding of the role of brain oscillations in language processing. Contributions from the pioneers of this field are selected, illustrating how the study of brain oscillations has allowed investigating theoretically relevant questions that could not be addressed by more traditional methods. The topic thus aims at deepening our mechanistic understanding of language processing and bringing us closer to bridging the gap between brain, mind and behavior for the crucial cognitive function of speech. > > Hosted By: > Marcela Pena, Catholic University of Chile, Chile > Lucia Melloni, Max Planck Institute for Brain Research, Germany -------------- next part -------------- An HTML attachment was scrubbed... URL: From Santosh.VemaKrishnamurthy at nrc-cnrc.gc.ca Mon Aug 15 14:06:43 2011 From: Santosh.VemaKrishnamurthy at nrc-cnrc.gc.ca (Vema Krishnamurthy, Santosh) Date: Mon, 15 Aug 2011 08:06:43 -0400 Subject: [FieldTrip] fieldtrip Digest, Vol 9, Issue 14 In-Reply-To: References: Message-ID: Hello Gregor, Thanks for the suggestions, I will give this a try. I'm not doing forward modeling with the high resolution surfaces. I'm trying to perform surface matching technique with scalp surface and the digitization data to do the co-registration of MEG and MRI. Thanks and Regards, --------------------------------------------------- Santosh Vema KrishnaMurthy, M.A.Sc Medical Devices Engineer NRC Institute for Biodiagnostics (Atlantic) Elekta/IWK MEG Research office Ph: (902) 470-3984 --------------------------------------------------- ________________________________________ From: fieldtrip-bounces at donders.ru.nl [fieldtrip-bounces at donders.ru.nl] On Behalf Of fieldtrip-request at donders.ru.nl [fieldtrip-request at donders.ru.nl] Sent: Saturday, August 13, 2011 7:00 AM To: fieldtrip at donders.ru.nl Subject: fieldtrip Digest, Vol 9, Issue 14 Send fieldtrip mailing list submissions to fieldtrip at donders.ru.nl To subscribe or unsubscribe via the World Wide Web, visit http://mailman.science.ru.nl/mailman/listinfo/fieldtrip or, via email, send a message with subject or body 'help' to fieldtrip-request at donders.ru.nl You can reach the person managing the list at fieldtrip-owner at donders.ru.nl When replying, please edit your Subject line so it is more specific than "Re: Contents of fieldtrip digest..." Today's Topics: 1. Scalp segmentation from an MRI (.nii or .fif) (Vema Krishnamurthy, Santosh) 2. Antw: Scalp segmentation from an MRI (.nii or .fif) (Gregor Volberg) ---------------------------------------------------------------------- Message: 1 Date: Fri, 12 Aug 2011 07:00:10 -0700 From: "Vema Krishnamurthy, Santosh" To: "fieldtrip at donders.ru.nl" Subject: [FieldTrip] Scalp segmentation from an MRI (.nii or .fif) Message-ID: Content-Type: text/plain; charset="iso-8859-1" Hello Everyone, I'm very new to Fieldtrip and I need to perform scalp segmentation from an mri file(.nii or .fif). I'm working on a project where I need the scalp surface for surface matching, so I need the scalp surface with high resolution which clearly shows the facial features (nose, eyes, lips, chin). Any help/suggestions relating to this are much appreciated. Thanks and Regards, --------------------------------------------------- Santosh Vema KrishnaMurthy, M.A.Sc Medical Devices Engineer NRC Institute for Biodiagnostics (Atlantic) Elekta/IWK MEG Research office Ph: (902) 470-3984 --------------------------------------------------- -------------- next part -------------- An HTML attachment was scrubbed... URL: ------------------------------ Message: 2 Date: Fri, 12 Aug 2011 18:47:30 +0200 From: "Gregor Volberg" To: Subject: [FieldTrip] Antw: Scalp segmentation from an MRI (.nii or .fif) Message-ID: <4E457543020000570000A631 at gwsmtp1.uni-regensburg.de> Content-Type: text/plain; charset="us-ascii" Dear Santosh, for that purpose you might want to have a look at the NFT toolbox from SCCN, http://sccn.ucsd.edu/nft/ . It does high resolution segmentations for brain, csf, scalp and skin surfaces. If you need the segmentation for computing forward solutions with BEM head models, then I would discourage a too fine resolution, though. It strongly increases the computational load. Best regards, Gregor -- Dr. rer. nat. Gregor Volberg ( mailto:gregor.volberg at psychologie.uni-regensburg.de ) University of Regensburg Institute for Experimental Psychology 93040 Regensburg, Germany Tel: +49 941 943 3862 Fax: +49 941 943 3233 http://www.psychologie.uni-regensburg.de/Greenlee/team/volberg/volberg.html >>> "Vema Krishnamurthy, Santosh" 12.08.11 16.42 Uhr >>> Hello Everyone, I'm very new to Fieldtrip and I need to perform scalp segmentation from an mri file(.nii or .fif). I'm working on a project where I need the scalp surface for surface matching, so I need the scalp surface with high resolution which clearly shows the facial features (nose, eyes, lips, chin). Any help/suggestions relating to this are much appreciated. Thanks and Regards, --------------------------------------------------- Santosh Vema KrishnaMurthy, M.A.Sc Medical Devices Engineer NRC Institute for Biodiagnostics (Atlantic) Elekta/IWK MEG Research office Ph: (902) 470-3984 --------------------------------------------------- -------------- next part -------------- An HTML attachment was scrubbed... URL: ------------------------------ _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip End of fieldtrip Digest, Vol 9, Issue 14 **************************************** From shreesb at yahoo.com Mon Aug 15 21:30:22 2011 From: shreesb at yahoo.com (shree b) Date: Mon, 15 Aug 2011 12:30:22 -0700 (PDT) Subject: [FieldTrip] Header file for TDT system Message-ID: <1313436622.23112.YahooMailClassic@web162012.mail.bf1.yahoo.com> Hello, I am new to field-trip and wanted to use your analysis on 16 channel continous LFP recordings from the TDT (Tucker Davis Technology) system. As there isn't a provision to read TDT header files I tried to input the values directly into the structure array. 1)I have already imported my data into Matlab and have the information that is need to be included in cfg.headerfile in a structure array: hdr=[]; hdr.Fs=1000; hdr.nChans=16; hdr.nSamples= abfi.dataPtsPerChan; %abfi.dataPtsPerChan=620646 hdr.nSamplesPre=0; hdr.nTrials=1; hdr.label= abfi.recChNames; %abfi.recChNames is a cell array with the channel names: %IN0,IN1,IN2,IN3,IN4,IN5,IN6,IN7,IN8,IN9,IN10,IN11,IN12,IN13,IN14,IN15 2) The data is in a mat file with all 16 channels labeled as before 3) I defined trl as: a=1; b=length(IN0); c=0; trl=[a b c];% single trial no offset cfg=struct('headerfile',hdr,'datafile', data,'trl',trl) This approach isn't working, could you please tell me how I should specify the values for the header file and data structure. thanks very much Shilpa From kirihara-tky at umin.ac.jp Tue Aug 16 06:00:11 2011 From: kirihara-tky at umin.ac.jp (Kenji Kirihara) Date: Tue, 16 Aug 2011 06:00:11 JST Subject: [FieldTrip] Time-frequency analysis Message-ID: <08160600.ME2066201@umin.org> Dear FieldTrip mailing list members, I am interested in time-frequency analysis of EEG and have a question about ft_freqanalysis. I applied ft_freqanalysis to my data for time-frequency analysis using wavelets. Power at each time-requency point is calculated. However, power at each point seems to be too large. For example, power at 4 Hz, 0 ms, O2 is 1218, power at 10 Hz is 730, and power at 20 Hz is 94. These values seem to be too large compared to power calculated using FFT for frequency analysis. I would be grateful if someone tells me why this happens. Sincerely, Kenji Kirihara From BelluscB at ninds.nih.gov Mon Aug 15 23:25:21 2011 From: BelluscB at ninds.nih.gov (Belluscio, Beth (NIH/NINDS) [E]) Date: Mon, 15 Aug 2011 17:25:21 -0400 Subject: [FieldTrip] preparing the forward model in beamforming Message-ID: <794DFDD3C128EF44AC49ADC58FD0090C057456B65B@NIHMLBX10.nih.gov> Hi Fieldtrip users- I am trying to learn to use the beamforming technique as outlined in the tutorial. I have a dicom MRI that I used to create a file that contains markers for the fiducials. I then used this in ft_read_mri and ft_volumesegment - both of which seemed to work fine. Then, when I used ft_prepare_singleshell, I get this error message: "??? Undefined function or method 'imfill' for input arguments of type 'char'. Error in ==> prepare_mesh_segmentation at 68 Seg = imfill((mri.seg==cfg.tisse(i)), 'holes')" I am used Fieldtrip 20110721 Any suggestions? Beth. Beth Belluscio MD-PhD Clinical Fellow Human Motor Control Section NINDS, NIH 301-402-3495 -------------- next part -------------- An HTML attachment was scrubbed... URL: From Don.Rojas at ucdenver.edu Mon Aug 15 23:57:07 2011 From: Don.Rojas at ucdenver.edu (Rojas, Don) Date: Mon, 15 Aug 2011 15:57:07 -0600 Subject: [FieldTrip] preparing the forward model in beamforming In-Reply-To: <794DFDD3C128EF44AC49ADC58FD0090C057456B65B@NIHMLBX10.nih.gov> References: <794DFDD3C128EF44AC49ADC58FD0090C057456B65B@NIHMLBX10.nih.gov> Message-ID: <2D825D9F-FEDA-4B1C-B401-38C3878B4E36@ucdenver.edu> Beth, The matlab function imfill.m appears to be missing for your matlab install. It is part of the matlab image processing toolbox, which is not part of the base matlab package. You could check with your local matlab admin to see if you should have the image processing toolbox. FieldTrip should probably check for the existence of this toolbox when used so that it can throw a more meaningful error message. Best, Don On Aug 15, 2011, at 3:25 PM, Belluscio, Beth (NIH/NINDS) [E] wrote: Hi Fieldtrip users- I am trying to learn to use the beamforming technique as outlined in the tutorial. I have a dicom MRI that I used to create a file that contains markers for the fiducials. I then used this in ft_read_mri and ft_volumesegment – both of which seemed to work fine. Then, when I used ft_prepare_singleshell, I get this error message: “??? Undefined function or method ‘imfill’ for input arguments of type ‘char’. Error in ==> prepare_mesh_segmentation at 68 Seg = imfill((mri.seg==cfg.tisse(i)), ‘holes’)” I am used Fieldtrip 20110721 Any suggestions? Beth. Beth Belluscio MD-PhD Clinical Fellow Human Motor Control Section NINDS, NIH 301-402-3495 _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip ----------------------- Don Rojas, Ph.D. Associate Professor of Psychiatry U. of Colorado Denver Anschutz Medical Campus Director, UCD Magnetoencephalography Lab 13001 E. 17th Pl F546 Aurora, CO 80045 303-724-4994 -------------- next part -------------- An HTML attachment was scrubbed... URL: From jan.schoffelen at donders.ru.nl Tue Aug 16 10:01:28 2011 From: jan.schoffelen at donders.ru.nl (jan-mathijs schoffelen) Date: Tue, 16 Aug 2011 10:01:28 +0200 Subject: [FieldTrip] Time-frequency analysis In-Reply-To: <08160600.ME2066201@umin.org> References: <08160600.ME2066201@umin.org> Message-ID: <5FE3FAAE-6A78-4033-B84C-AEB55E844285@donders.ru.nl> Dear Kenji, In addition of applying the fft on the data, fieldtrip normalises the fft with a number that equals 2/Nsmp, where Nsmp is the length of the wavelet. Best wishes, Jan-Mathijs On Aug 15, 2011, at 11:00 PM, Kenji Kirihara wrote: > Dear FieldTrip mailing list members, > > I am interested in time-frequency analysis of EEG and have > a question about ft_freqanalysis. I applied ft_freqanalysis to > my data for time-frequency analysis using wavelets. Power at each > time-requency point is calculated. However, power at each point > seems to be too large. For example, power at 4 Hz, 0 ms, O2 is 1218, > power at 10 Hz is 730, and power at 20 Hz is 94. These values seem > to be too large compared to power calculated using FFT for frequency > analysis. I would be grateful if someone tells me why this happens. > > Sincerely, > > Kenji Kirihara > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip Dr. J.M. (Jan-Mathijs) Schoffelen Donders Institute for Brain, Cognition and Behaviour, Centre for Cognitive Neuroimaging, Radboud University Nijmegen, The Netherlands J.Schoffelen at donders.ru.nl Telephone: +31-24-3614793 From eelke.spaak at donders.ru.nl Tue Aug 16 10:36:16 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Tue, 16 Aug 2011 10:36:16 +0200 Subject: [FieldTrip] Header file for TDT system In-Reply-To: <1313436622.23112.YahooMailClassic@web162012.mail.bf1.yahoo.com> References: <1313436622.23112.YahooMailClassic@web162012.mail.bf1.yahoo.com> Message-ID: Dear Shilpa, If you already have the data and related information in Matlab, it is probably best to skip the entire trial-definition step when converting your data to FieldTrip format. It's quite straightforward to convert data into a format that FT can handle, especially if you just want it to have a single segment (or 'trial'). In your case, from the top of my head, you could initialize a structure with the following fields: data.label = abfi.recChNames; data.trial = {yourDataMatrix}; % assuming size(yourDataMatrix) == [nChans nSamples] data.fsample = 1000; data.time = {0:1/data.fsample:lastTimePoint}; data.sampleinfo = [1 nSamples]; This is just about the bare minimum that FT requires, which in your case should be sufficient. Note that both the data.trial and data.time fields should be cell arrays, with one element per trial in your data. For more detailed information about how FT raw data should look, see the following entry on our wiki: http://fieldtrip.fcdonders.nl/faq/how_are_the_various_data_structures_defined . Hope this was of help, Best, Eelke 2011/8/15 shree b : > Hello, > I am new to field-trip and wanted to use your analysis on 16 channel continous LFP recordings from the TDT (Tucker Davis Technology) system. As there isn't a provision to read TDT header files I tried to input the values directly into the structure array. > > 1)I have already imported my data into Matlab and have the information that is need to be included in cfg.headerfile in a structure array: > > hdr=[]; > > hdr.Fs=1000; > hdr.nChans=16; > hdr.nSamples= abfi.dataPtsPerChan; %abfi.dataPtsPerChan=620646 > hdr.nSamplesPre=0; > hdr.nTrials=1; > hdr.label= abfi.recChNames; > %abfi.recChNames is a cell array with the channel names:    %IN0,IN1,IN2,IN3,IN4,IN5,IN6,IN7,IN8,IN9,IN10,IN11,IN12,IN13,IN14,IN15 > > 2) The data is in a mat file with all 16 channels labeled as before > > 3) I defined trl as: > a=1; > b=length(IN0); > c=0; > > trl=[a b c];% single trial no offset > > cfg=struct('headerfile',hdr,'datafile', data,'trl',trl) > > This approach isn't working, could you please tell me how I should specify the values for the header file and data structure. > > thanks very much > Shilpa > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From jakobala at hotmail.com Tue Aug 16 14:09:05 2011 From: jakobala at hotmail.com (jakob kaiser) Date: Tue, 16 Aug 2011 14:09:05 +0200 Subject: [FieldTrip] scaling frequency data for comparison Message-ID: Dear list,I wanted to compare the outputs from different ft_freqanalysis-methods to see if they make significant differences in the final test results of my data. This were the parameters used: cfg.output = 'pow';cfg.channel = 'all';cfg.method = 'mtmconvol';cfg.taper = 'hanning';cfg.foi = 6:2:20; cfg.t_ftimwin = ones(length(cfg.foi),1).*0.5; cfg.toi = -0.5:0.05:2; WaveletConfig.method = 'wavelet'; WaveletConfig.output = 'pow'; WaveletConfig.foi = 6:2:20; WaveletConfig.toi = -0.5:0.05:2; (I chose these two methods because I found them used un papers relevant for my work)Now, as has been discussed on the list before, the scaling of the output data is very different. While the mtconvol-hanning-method gives me data points between roughly -2 an 1, with the wavelet-method I have data between roughly -800 and 600. From the (btw very helpful) tutorials and the previous list discussions I understand that this is a normal byproduct of the mathematical methods employed. However, for comparing different outputs it would be useful to scale both produced data sets to the same unit, so to speak. As both methods are supposed to measure the same thing, it should someone be possible to scale them to similar sizes, shouldn't it? Is there such a way? (Ultimately, I will have to compare the data results to results from a completely different program, which seem to have a third, different scaling altogether, so I am really puzzled how to handle the differences in scale)Generally, although this has been mentioned before, I am still not sure, if it is possible to put a certain unit to the results. Power is often reported in mV^2/Hz. If my initial data is in mV, would it be correct to say, that the hanning-method describe above delivers data in mV^2/Hz? Would this be correct for the Wavelet-method? Obviously, both cannot be true, because the scaling is extremely different. So, is it possible to specify a unit for any of these methods? Is there a way to convert the output of ft_freqanalysis-methods to this or another meaningful unit?I would be very grateful, if someone could give me a hint, how to interpret the data points here.Thank you for reading + thanks to the programmers for fieldtrip, which I like a lot Jakob -------------- next part -------------- An HTML attachment was scrubbed... URL: From eelke.spaak at donders.ru.nl Tue Aug 16 14:20:45 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Tue, 16 Aug 2011 14:20:45 +0200 Subject: [FieldTrip] scaling frequency data for comparison In-Reply-To: References: Message-ID: Dear Jakob, I can't give you any definitive answer on the power units of the different freqanalysis algorithms, but you might find the function ft_freqbaseline helpful. Absolute power values are usually not very meaningful in neuroscience research, since they depend on a lot of non-brain factors. Ft_freqbaseline performs a baseline correction (either relative, absolute, or percentage) on your power values. Across different spectral estimation methods, the relative power change should remain the same (since it is unitless). Best, Eelke 2011/8/16 jakob kaiser : > Dear list, > I wanted to compare the outputs from different ft_freqanalysis-methods to > see if they make significant differences in the final test results of my > data. This were the parameters used: > > cfg.output       = 'pow'; > cfg.channel      = 'all'; > cfg.method       = 'mtmconvol'; > cfg.taper        = 'hanning'; > cfg.foi          = 6:2:20; > cfg.t_ftimwin    = ones(length(cfg.foi),1).*0.5; > cfg.toi          = -0.5:0.05:2; > > WaveletConfig.method = 'wavelet'; > WaveletConfig.output = 'pow'; > WaveletConfig.foi = 6:2:20; > WaveletConfig.toi = -0.5:0.05:2; > (I chose these two methods because I found them used un papers relevant for > my work) > Now, as has been discussed on the list before, the scaling of the output > data is very different. While the mtconvol-hanning-method gives me data > points between roughly -2 an 1, with the wavelet-method I have data between > roughly -800 and 600. From the (btw very helpful) tutorials and the previous > list discussions I understand that this is a normal byproduct of > the mathematical methods employed. However, for comparing different outputs > it would be useful to scale both produced data sets to the same unit, so to > speak. As both methods are supposed to measure the same thing, it should > someone be possible to scale them to similar sizes, shouldn't it? Is there > such a way? (Ultimately, I will have to compare the data results to results > from a completely different program, which seem to have a third, different > scaling altogether, so I am really puzzled how to handle the differences in > scale) > Generally, although this has been mentioned before, I am still not sure, if > it is possible to put a certain unit to the results. Power is often reported > in mV^2/Hz. If my initial data is in mV, would it be correct to say, that > the hanning-method describe above delivers data in mV^2/Hz? Would this be > correct for the Wavelet-method? Obviously, both cannot be true, because the > scaling is extremely different. So, is it possible to specify a unit for any > of these methods? Is there a way to convert the output > of ft_freqanalysis-methods to this or another meaningful unit? > I would be very grateful, if someone could give me a hint, how to interpret > the data points here. > Thank you for reading + thanks to the programmers for fieldtrip, which I > like a lot > Jakob > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From kirihara-tky at umin.ac.jp Wed Aug 17 02:33:15 2011 From: kirihara-tky at umin.ac.jp (Kenji Kirihara) Date: Wed, 17 Aug 2011 02:33:15 JST Subject: [FieldTrip] Time-frequency analysis Message-ID: <08170233.ME2758201@umin.org> Dear Jan-Mathijs, Thank you for your advice. Sincerely, Kenji, In message <5FE3FAAE-6A78-4033-B84C-AEB55E844285 at donders.ru.nl> fieldtrip at donder s.ru.nl (jan-mathijs schoffelen) wrote: > From: jan-mathijs schoffelen > Date: Tue, 16 Aug 2011 10:01:28 +0200 > To: Email discussion list for the FieldTrip project > Subject: Re: [FieldTrip] Time-frequency analysis > > Dear Kenji, > > In addition of applying the fft on the data, fieldtrip normalises the fft with a number that equals 2/Nsmp, where Nsmp is the length of the wavelet. > > Best wishes, > > Jan-Mathijs > > > On Aug 15, 2011, at 11:00 PM, Kenji Kirihara wrote: > > > Dear FieldTrip mailing list members, > > > > I am interested in time-frequency analysis of EEG and have > > a question about ft_freqanalysis. I applied ft_freqanalysis to > > my data for time-frequency analysis using wavelets. Power at each > > time-requency point is calculated. However, power at each point > > seems to be too large. For example, power at 4 Hz, 0 ms, O2 is 1218, > > power at 10 Hz is 730, and power at 20 Hz is 94. These values seem > > to be too large compared to power calculated using FFT for frequency > > analysis. I would be grateful if someone tells me why this happens. > > > > Sincerely, > > > > Kenji Kirihara > > _______________________________________________ > > fieldtrip mailing list > > fieldtrip at donders.ru.nl > > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > Dr. J.M. (Jan-Mathijs) Schoffelen > Donders Institute for Brain, Cognition and Behaviour, > Centre for Cognitive Neuroimaging, > Radboud University Nijmegen, The Netherlands > J.Schoffelen at donders.ru.nl > Telephone: +31-24-3614793 > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From g.piantoni at nin.knaw.nl Wed Aug 17 12:43:23 2011 From: g.piantoni at nin.knaw.nl (Gio Piantoni) Date: Wed, 17 Aug 2011 12:43:23 +0200 Subject: [FieldTrip] Granger convention Message-ID: Hi all, The connectivity toolbox is very powerful and straightforward to use, thanks for that! However, I did not find any information about the convention used by Fieldtrip to represent asymmetrical connectivity measures (such as granger causality), f.e. in the help of the functions, in http://fieldtrip.fcdonders.nl/faq/in_what_way_can_frequency_domain_data_be_represented_in_fieldtrip nor in http://fieldtrip.fcdonders.nl/development/connectivity_tutorial Do the values in data.grangerspctrm(1, 2, :) represent the granger-causality values from channel 1 to channel 2 or the other way round? BTW, is it possible to run time-domain granger causality in Fieldtrip? Thanks! Gio -- Giovanni Piantoni, Ph.D. student Dept. Sleep & Cognition Netherlands Institute for Neuroscience Meibergdreef 47 1105 BA Amsterdam (NL) +31 (0)20 5665492 g.piantoni at nin.knaw.nl www.giovannipiantoni.com From jpnv2006 at gmail.com Wed Aug 17 14:58:07 2011 From: jpnv2006 at gmail.com (Juan Pablo Neira) Date: Wed, 17 Aug 2011 14:58:07 +0200 Subject: [FieldTrip] Forward solution using concentric spheres and BEM models Message-ID: Hello, I am working with an individual MRI and EEG data to do the forward solution with two different volume's model (concentric spheres and BEM), so at the end i can compare the results.  But i did not have the expected results of the concentric spheres model until now. This is my script: 1. Read individual MRI mri=ft_read_mri('data_patient_1\******.hdr'); %I got this "Warning: flipping 1st dimension (L-R) to obtain volume in neurological convention" %How can i do a homogenous transformation matrix, using the voxel dimensions %that are specified in hdr.dime.pixdim? 2. Realign to 'head coordinates' 3. MRI Reslicing 4. MRI segmentation: (brain, skull,scalp) 5. Create geometrical description of the brain, skull and scalp (3 compartments) cfg = []; cfg.interactive = 'no'; %segmentation method cfg.numvertices = 3000; cfg.sourceunits = 'mm'; cfg.downsample = 2; cfg.numcompartments = 3; cfg.tissue = {'brain' 'skull' 'scalp'}; cfg.smooth = 'yes'; bnd = ft_prepare_mesh(cfg, mri_segment); %The output just give me the geometrical description of the brain. How can i get the geometrical description of %the skull and scalp also to proceed to create a 3 spheres concentric sphere model? 6. If i have the 3 geometrical descriptions (brain, skull and scalp). Create a 3 concentric spheres model. cfg = []; cfg.headshape = [bnd1 bnd2 bnd3]; % brain- skull - scalp cfg.conductivity = [0.33 0.0042 0.331]; [vol, cfg] = prepare_concentricspheres(cfg); %I got the model but part of the brain is outside of its sphere and also the skull. How can I increase the radius so % the whole brain will be inside its sphere and also the skull? The sphere of the scalp fix good. I hope someone can help me solve these questions. Regards, Juan Pablo From sysoevao at psychiatry.wustl.edu Thu Aug 18 05:39:11 2011 From: sysoevao at psychiatry.wustl.edu (Sysoeva, Olga Vladimirovna) Date: Wed, 17 Aug 2011 22:39:11 -0500 Subject: [FieldTrip] mixed between/within subject Anova Message-ID: <76DC57362137854CB9592B496182D5FA0506E253FF@wusmexmbx2.medpriv.wucon.wustl.edu> Dear fieldtrip list members, I'd like to use mixed between subjects (3 groups) and within subject (2 factors each has 2 levels) Anova design to compare the ERPs using cluster permutation statistics (ft_timelockananlysis). But to my understanding it is impossible. Am I right? You comments are highly appreciated, Best Regards, Olga. Olga Sysoeva, Research Associate, PhD Washington University School of Medicine Campus Box 8134 660 South Euclid Ave Saint Louis, MO 63110-9909 The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. From sysoevao at psychiatry.wustl.edu Thu Aug 18 05:41:36 2011 From: sysoevao at psychiatry.wustl.edu (Sysoeva, Olga Vladimirovna) Date: Wed, 17 Aug 2011 22:41:36 -0500 Subject: [FieldTrip] cluster analysis for peak/latency data Message-ID: <76DC57362137854CB9592B496182D5FA0506E253FD@wusmexmbx2.medpriv.wucon.wustl.edu> Dear fieldtrip list members, I’m applying cluster permutation algorithm, implicated in Fieldtrip software to my EEG data. But I’m using it in a little bit different way. I want to cluster my data from 32 channels for peak amplitude and latency measures, not raw ERP data. So, as an input I have only chn*subjects matrix, without time dimension. Technically the script works fine (ft_timelockstatistics), and, from my point of view, it also makes perfect theoretic sense: we can examined both peak and latency of the component ( in point by point comparison it might be confusing to disentangle these effects). Certainly, the main motivation for this cluster-permutation test was to deal with multiple comparison problem and to conduct exploratory investigation (huge datasets). But in case of looking a specific component cluster permutation analysis is helping to deal with MCP introduced by multiple channels. Am I right? You comments are highly appreciated, Best Regards, Olga. Olga Sysoeva, Research Associate, PhD Washington University School of Medicine Campus Box 8134 660 South Euclid Ave Saint Louis, MO 63110-9909 The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. From e.maris at donders.ru.nl Thu Aug 18 07:40:32 2011 From: e.maris at donders.ru.nl (Eric Maris) Date: Thu, 18 Aug 2011 07:40:32 +0200 Subject: [FieldTrip] cluster analysis for peak/latency data In-Reply-To: <76DC57362137854CB9592B496182D5FA0506E253FD@wusmexmbx2.medpriv.wucon.wustl.edu> References: <76DC57362137854CB9592B496182D5FA0506E253FD@wusmexmbx2.medpriv.wucon.wustl.edu> Message-ID: <024f01cc5d69$58e406f0$0aac14d0$@maris@donders.ru.nl> Dear Olga, > But I'm using it in a little bit different way. I want to cluster my data from 32 > channels for peak amplitude and latency measures, not raw ERP data. So, as > an input I have only chn*subjects matrix, without time dimension. > Technically the script works fine (ft_timelockstatistics), and, from my point > of view, it also makes perfect theoretic sense: we can examined both peak > and latency of the component ( in point by point comparison it might be > confusing to disentangle these effects). Certainly, the main motivation for > this cluster-permutation test was to deal with multiple comparison problem > and to conduct exploratory investigation (huge datasets). But in case of > looking a specific component cluster permutation analysis is helping to deal > with MCP introduced by multiple channels. Am I right? If you do this separately for peak latency and peak amplitude, I see no problem with this approach. Best, Eric > > You comments are highly appreciated, > > Best Regards, > Olga. > > > > Olga Sysoeva, > Research Associate, PhD > Washington University School of Medicine > Campus Box 8134 > 660 South Euclid Ave > Saint Louis, MO 63110-9909 > > The materials in this message are private and may contain Protected > Healthcare Information or other information of a sensitive nature. If you are > not the intended recipient, be advised that any unauthorized use, disclosure, > copying or the taking of any action in reliance on the contents of this > information is strictly prohibited. If you have received this email in error, > please immediately notify the sender via telephone or return mail. > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From e.maris at donders.ru.nl Thu Aug 18 07:50:08 2011 From: e.maris at donders.ru.nl (Eric Maris) Date: Thu, 18 Aug 2011 07:50:08 +0200 Subject: [FieldTrip] mixed between/within subject Anova In-Reply-To: <76DC57362137854CB9592B496182D5FA0506E253FF@wusmexmbx2.medpriv.wucon.wustl.edu> References: <76DC57362137854CB9592B496182D5FA0506E253FF@wusmexmbx2.medpriv.wucon.wustl.edu> Message-ID: <025001cc5d6a$b00dc230$10294690$@maris@donders.ru.nl> Dear Olga, > I'd like to use mixed between subjects (3 groups) and within subject (2 > factors each has 2 levels) Anova design to compare the ERPs using cluster > permutation statistics (ft_timelockananlysis). > But to my understanding it is impossible. Am I right? The problem is with the interaction effects. With permutation tests, you cannot test the interaction between two (or more) between-UO variables (UO=subject or trial) nor the interaction between two or more within-UO variables. However, with permutation tests, you can test the interaction between a between- and a within-UO variables. So. In your study, you would only be unable to test the interaction between the two within-subject variables. There have been more threads on the testing of interaction effects. You can find them by searching in the discussion list archive. Best, Eric Maris > > You comments are highly appreciated, > > Best Regards, > Olga. > > > Olga Sysoeva, > Research Associate, PhD > Washington University School of Medicine > Campus Box 8134 > 660 South Euclid Ave > Saint Louis, MO 63110-9909 > > The materials in this message are private and may contain Protected > Healthcare Information or other information of a sensitive nature. If you are > not the intended recipient, be advised that any unauthorized use, disclosure, > copying or the taking of any action in reliance on the contents of this > information is strictly prohibited. If you have received this email in error, > please immediately notify the sender via telephone or return mail. > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From michael.wibral at web.de Thu Aug 18 09:57:03 2011 From: michael.wibral at web.de (Michael Wibral) Date: Thu, 18 Aug 2011 09:57:03 +0200 (CEST) Subject: [FieldTrip] mixed between/within subject Anova Message-ID: <961614915.5812596.1313654223009.JavaMail.fmail@mwmweb069> Dear Olga, you might find this reference helpful with respect to permutation tests and ANOVA-like questions: PERMUTATION TESTS FOR MULTI-FACTORIAL ANALYSIS OF VARIANCE MARTI J. ANDERSON and CAJO J. F. TER BRAAK Journal of Statistical Computation and Simulation, 2003, Vol. 73(2), pp. 85–113 Michael (let me know if you have trouble getting it) -----Ursprüngliche Nachricht----- Von: "Eric Maris" Gesendet: Aug 18, 2011 7:50:08 AM An: "'Email discussion list for the FieldTrip project'" Betreff: Re: [FieldTrip] mixed between/within subject Anova >Dear Olga, > >> I'd like to use mixed between subjects (3 groups) and within subject (2 >> factors each has 2 levels) Anova design to compare the ERPs using cluster >> permutation statistics (ft_timelockananlysis). >> But to my understanding it is impossible. Am I right? > >The problem is with the interaction effects. With permutation tests, you >cannot test the interaction between two (or more) between-UO variables >(UO=subject or trial) nor the interaction between two or more within-UO >variables. However, with permutation tests, you can test the interaction >between a between- and a within-UO variables. So. In your study, you would >only be unable to test the interaction between the two within-subject >variables. > >There have been more threads on the testing of interaction effects. You can >find them by searching in the discussion list archive. > >Best, > >Eric Maris > > > > >> >> You comments are highly appreciated, >> >> Best Regards, >> Olga. >> >> >> Olga Sysoeva, >> Research Associate, PhD >> Washington University School of Medicine >> Campus Box 8134 >> 660 South Euclid Ave >> Saint Louis, MO 63110-9909 >> >> The materials in this message are private and may contain Protected >> Healthcare Information or other information of a sensitive nature. If you >are >> not the intended recipient, be advised that any unauthorized use, >disclosure, >> copying or the taking of any action in reliance on the contents of this >> information is strictly prohibited. If you have received this email in >error, >> please immediately notify the sender via telephone or return mail. >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > >_______________________________________________ >fieldtrip mailing list >fieldtrip at donders.ru.nl >http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From Margit.Schoenherr at uk-erlangen.de Thu Aug 18 11:15:25 2011 From: Margit.Schoenherr at uk-erlangen.de (=?iso-8859-1?Q?Sch=F6nherr=2C_Margit?=) Date: Thu, 18 Aug 2011 11:15:25 +0200 Subject: [FieldTrip] Forward solution using concentric spheres and BEM models In-Reply-To: References: Message-ID: <71D8F8A56F37A947912FC86D4D9F00F4633F43AE27@XMAIL1.medads.uk-erlangen.de> Hello, I probably cannot help you with all your questions, but currently I'm also dealing with MRI segmentation and volume conductors, so maybe my solutions can be useful for you. Concerning your point 5 (geometrical description of the brain, skull, and scalp), I have done it the following way: cfg = []; cfg.method = 'segmentation'; cfg.tissue = [1 2 3]; % scalp = 1; skull = 2; brain = 3; cfg.numvertices = [2000 1000 800]; cfg.sourceunits = 'mm'; cfg.mriunits = 'mm'; bnd = ft_prepare_mesh(cfg, mri_segment); I think you have to specify a vector 'cfg.numvertices', a number of vertices for each compartment. By the way, at the beginning of ft_prepare_mesh the cfg-field 'numcompartments' is removed from the cfg structure (at least in my fieldtrip version, which is not the latest). Another question that comes to me is, why you need the segmented MRI for a concentric spheres volume conductor? So, regarding point 6, you can build such a vol-structure for example like this: vol.unit = 'm'; vol.o = origin; vol.r = [R_brain R_skull R_skin]; vol.c = [1 1/80 1]; By this, you can also define the radius as you like. Best, Margit ________________________________________ Von: fieldtrip-bounces at donders.ru.nl [fieldtrip-bounces at donders.ru.nl] im Auftrag von Juan Pablo Neira [jpnv2006 at gmail.com] Gesendet: Mittwoch, 17. August 2011 14:58 An: fieldtrip at donders.ru.nl Betreff: [FieldTrip] Forward solution using concentric spheres and BEM models Hello, I am working with an individual MRI and EEG data to do the forward solution with two different volume's model (concentric spheres and BEM), so at the end i can compare the results. But i did not have the expected results of the concentric spheres model until now. This is my script: 1. Read individual MRI mri=ft_read_mri('data_patient_1\******.hdr'); %I got this "Warning: flipping 1st dimension (L-R) to obtain volume in neurological convention" %How can i do a homogenous transformation matrix, using the voxel dimensions %that are specified in hdr.dime.pixdim? 2. Realign to 'head coordinates' 3. MRI Reslicing 4. MRI segmentation: (brain, skull,scalp) 5. Create geometrical description of the brain, skull and scalp (3 compartments) cfg = []; cfg.interactive = 'no'; %segmentation method cfg.numvertices = 3000; cfg.sourceunits = 'mm'; cfg.downsample = 2; cfg.numcompartments = 3; cfg.tissue = {'brain' 'skull' 'scalp'}; cfg.smooth = 'yes'; bnd = ft_prepare_mesh(cfg, mri_segment); %The output just give me the geometrical description of the brain. How can i get the geometrical description of %the skull and scalp also to proceed to create a 3 spheres concentric sphere model? 6. If i have the 3 geometrical descriptions (brain, skull and scalp). Create a 3 concentric spheres model. cfg = []; cfg.headshape = [bnd1 bnd2 bnd3]; % brain- skull - scalp cfg.conductivity = [0.33 0.0042 0.331]; [vol, cfg] = prepare_concentricspheres(cfg); %I got the model but part of the brain is outside of its sphere and also the skull. How can I increase the radius so % the whole brain will be inside its sphere and also the skull? The sphere of the scalp fix good. I hope someone can help me solve these questions. Regards, Juan Pablo _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From litvak.vladimir at gmail.com Thu Aug 18 11:37:59 2011 From: litvak.vladimir at gmail.com (Vladimir Litvak) Date: Thu, 18 Aug 2011 10:37:59 +0100 Subject: [FieldTrip] bdf files persisting in memory? In-Reply-To: <028001cc5d3f$45e3edb0$d1abc910$%Budd@newcastle.edu.au> References: <9231708890694273.WA.cl304kent.ac.uk@jiscmail.ac.uk> <028001cc5d3f$45e3edb0$d1abc910$%Budd@newcastle.edu.au> Message-ID: Dear Bill, This might be a memory leek in the low-level mex file. It seems that the code for this file is available so it can be fixed. I'm CCing this to the Fieldtrip list and I'll also submit a bug report to Fieldtrip developers. Best, Vladimir On Thu, Aug 18, 2011 at 1:39 AM, Bill Budd wrote: > Dear Vladimir > > I've noticed when using spm_eeg_convert to convert bdf files that the bdf > file persists in memory after the spm_eeg_convert function or even the > script calling it has finished. Not sure if the issue is specific to my > scripting or the spm/fieldtrip functions. Wondered if you have any advice as > it uses a lot of memory while the rest of the script pipeline runs. The only > way I can seem to clear it is to run 'clear all mex' or exit matlab, but > really need some less sledge-hammer style solution to clearing it from RAM. > > Cheers > -Bill > > From drivolta81 at gmail.com Thu Aug 18 15:18:19 2011 From: drivolta81 at gmail.com (Davide Rivolta) Date: Thu, 18 Aug 2011 15:18:19 +0200 Subject: [FieldTrip] common filters - beamforming Message-ID: Dear Fieldtrippers, I am trying to learn beamforming using the tutorial. I am succesful in calculating, with my own data, the relative change between the stuimulus and the baseline condition (as shown in the tutorial). However, if I want to do statistic (ie., compare the baseline and stimulus condition) I would need a common filter (not in the tutorial). I found very useful the website where this step is explained (examples of matlab scripts section). I am however confused on how to perform stats at the source level. I get the 2 conditions separately and I wish to compare them and hopefully plot WHERE the difference is seen. Do you have any advice on how to do it (or where to look)? Here is my script and I am sure I am doing some silly mistake in the last (statistic part) since it gives the error: ??? Reference to non-existent field 'pow'. cfg = []; cfg.toilim = [-0.5 0]; dataPre = ft_redefinetrial(cfg, DataOut); cfg.toilim = [0.5 1]; dataPost = ft_redefinetrial(cfg, DataOut); data = ft_appenddata([], dataPre, dataPost); design = [ones(1, length(dataPre.trial)) ones(1, length(dataPost.trial))*2]; cfg = []; cfg.method = 'mtmfft'; cfg.output = 'powandcsd'; cfg.taper = 'dpss'; cfg.keeptrials = 'yes'; cfg.keeptapers = 'no'; cfg.foi = 57:2:63; cfg.tapsmofrq = ones(length(cfg.foi), 1).* 5; freq = ft_freqanalysis(cfg, data); cfg = []; cfg.frequency = 60; cfg.method = 'dics'; cfg.grid = grid; cfg.vol = vol; cfg.grad = DataOut.hdr.grad; cfg.keepfilter = 'yes'; source = ft_sourceanalysis(cfg, freq); cfg = []; cfg.frequency = 60; cfg.method = 'dics'; cfg.grid = grid; cfg.vol = vol; cfg.grad = DataOut.hdr.grad; cfg.grid.filter = source.avg.filter; %uses a precomputed filter (common filter) cfg.rawtrial = 'yes'; source = ft_sourceanalysis(cfg, freq); A = find(design==1); B = find(design==2); sourceA = source; sourceA.trial(B) = []; sourceA.cumtapcnt(B) = []; sourceA.df = length(A); cfg = []; cfg.keeptrials = 'yes'; sourceA = ft_sourcedescriptives([], sourceA); sourceB = source; sourceB.trial(A) = []; sourceB.cumtapcnt(A) = []; sourceB.df = length(B); cfg = []; cfg.keeptrials = 'yes'; sourceB = ft_sourcedescriptives([], sourceB); %STATS cfg = []; cfg.parameter = 'pow'; cfg.method = 'analytic'; cfg.statistic = 'actvsblT'; cfg.computestat = 'yes'; cfg.computecritval = 'yes'; cfg.computeprob = 'yes'; cfg.alpha = 0.05; cfg.tail = 0; stat = ft_sourcestatistics(cfg, sourceB, sourceA); Thanks for your help, Davide -------------- next part -------------- An HTML attachment was scrubbed... URL: From sysoevao at psychiatry.wustl.edu Thu Aug 18 18:02:11 2011 From: sysoevao at psychiatry.wustl.edu (Sysoeva, Olga Vladimirovna) Date: Thu, 18 Aug 2011 11:02:11 -0500 Subject: [FieldTrip] mixed between/within subject Anova In-Reply-To: <961614915.5812596.1313654223009.JavaMail.fmail@mwmweb069> References: <961614915.5812596.1313654223009.JavaMail.fmail@mwmweb069> Message-ID: <76DC57362137854CB9592B496182D5FA0506F6EAFB@wusmexmbx2.medpriv.wucon.wustl.edu> Thank you Michael, Thank you Eric Maris, So, as far as I understand there are some suggestion (Anderson et al, 2003) about how to apply the permutation test to my case (1 between and 2 within subjects factors), but it is not implemented in fieldtrip yet. As for examination the interaction for 1 between and 1 within subject factors - it is possible to do with fieldtrip scripts, isn't it? I've looked through the archive and one of the option seemed to be just to examine the between subject effect on the data obtained from subtraction of one level of within subject variable from the another one ( in case of 2 levels). And the results will represent the interaction effect. Am I right? Best Regards, Olga. -----Original Message----- From: fieldtrip-bounces at donders.ru.nl [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Michael Wibral Sent: Thursday, August 18, 2011 2:57 AM To: Email discussion list for the FieldTrip project Subject: Re: [FieldTrip] mixed between/within subject Anova Dear Olga, you might find this reference helpful with respect to permutation tests and ANOVA-like questions: PERMUTATION TESTS FOR MULTI-FACTORIAL ANALYSIS OF VARIANCE MARTI J. ANDERSON and CAJO J. F. TER BRAAK Journal of Statistical Computation and Simulation, 2003, Vol. 73(2), pp. 85–113 Michael (let me know if you have trouble getting it) -----Ursprüngliche Nachricht----- Von: "Eric Maris" Gesendet: Aug 18, 2011 7:50:08 AM An: "'Email discussion list for the FieldTrip project'" Betreff: Re: [FieldTrip] mixed between/within subject Anova >Dear Olga, > >> I'd like to use mixed between subjects (3 groups) and within subject >> (2 factors each has 2 levels) Anova design to compare the ERPs using >> cluster permutation statistics (ft_timelockananlysis). >> But to my understanding it is impossible. Am I right? > >The problem is with the interaction effects. With permutation tests, >you cannot test the interaction between two (or more) between-UO >variables (UO=subject or trial) nor the interaction between two or more >within-UO variables. However, with permutation tests, you can test the >interaction between a between- and a within-UO variables. So. In your >study, you would only be unable to test the interaction between the two >within-subject variables. > >There have been more threads on the testing of interaction effects. You >can find them by searching in the discussion list archive. > >Best, > >Eric Maris > > > > >> >> You comments are highly appreciated, >> >> Best Regards, >> Olga. >> >> >> Olga Sysoeva, >> Research Associate, PhD >> Washington University School of Medicine Campus Box 8134 >> 660 South Euclid Ave >> Saint Louis, MO 63110-9909 >> >> The materials in this message are private and may contain Protected >> Healthcare Information or other information of a sensitive nature. If >> you >are >> not the intended recipient, be advised that any unauthorized use, >disclosure, >> copying or the taking of any action in reliance on the contents of >> this information is strictly prohibited. If you have received this >> email in >error, >> please immediately notify the sender via telephone or return mail. >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > >_______________________________________________ >fieldtrip mailing list >fieldtrip at donders.ru.nl >http://mailman.science.ru.nl/mailman/listinfo/fieldtrip _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. From sust2001 at yahoo.ca Thu Aug 18 21:22:56 2011 From: sust2001 at yahoo.ca (Mark Taylor) Date: Thu, 18 Aug 2011 12:22:56 -0700 (PDT) Subject: [FieldTrip] Need your help to run *.ntt data file in the FieldTrip Message-ID: <1313695376.13817.YahooMailNeo@web46410.mail.sp1.yahoo.com> Hi, Could you please let me know whether I can  run the spike data file recorded with *.ntt format from Neuralynx in the FieldTrip environment? As this *.ntt format is not listed in the option of supported spike and LFP data formats, I would be very grateful if I can get your valuable suggestions to make the FieldTrip useful for my recorded data analysis. Your kind help and the earliest reply would be really appreciated. Thanking you, Mark. -------------- next part -------------- An HTML attachment was scrubbed... URL: From Ulrich.Pomper at charite.de Fri Aug 19 12:27:20 2011 From: Ulrich.Pomper at charite.de (Pomper, Ulrich) Date: Fri, 19 Aug 2011 12:27:20 +0200 Subject: [FieldTrip] =?windows-1252?q?1_PhD_and_1_Postdoctoral_Position_-_?= =?windows-1252?q?Charit=E9_Berlin?= Message-ID: The Department of Psychiatry and Psychotherapy, Charité – Universitätsmedizin Berlin (http://psy-ccm.charite.de/en/) invites applications for a Post-doctoral and a PhD student position. A Starting Grant of the European Research Council (ERC) will fund both positions for initially 2 years with the possibility of extension. The main objective of this ERC research program is to examine neural and neurochemical markers of multisensory integration and to test a new hypothesis that considers dynamic interplay of synchronized neural populations as a key to multisensory processes (Senkowski et al. 2008, Trends in Neurosciences). The studies within this program include healthy subjects and patients with schizophrenia, as a prototype of a mental disorder with deficits in multisensory integration. Multisensory processes will be examined in a series of experiments requiring both bottom-up and top-down processing (Talsma et al. 2010, Trends in Cognitive Sciences). Reaching beyond the state-of-the-art, this comprises a combination of human EEG/MEG data as a macroscopic measure of cortical processing; source modeling of synchronized EEG/MEG activity; and neurochemical markers using MR spectroscopy. Applicants should have a background in psychology, medicine, biology, physics or neuroscience. Experience in human EEG/MEG studies or biosignal analysis (e.g., Matlab) is desirable (i.e. required for the Post-doctoral position). An interest in neurophysiologic studies in clinical populations is expected, as well as basic German language skills for interacting with patients. Applicants are asked to submit their CV, a short motivation letter, 2 names of referees, and documentation of relevant qualification (e.g., copies of diplomas and/or transcripts of grades) until September 15, 2011, electronically to Dr. Daniel Senkowski (daniel.senkowski at charite.de; www.danielsenkowski.com). --------------------------------------------------------------- Daniel Senkowski, Ph.D Head of the research group Multisensory-Mind Department of Psychiatry and Psychotherapy Charité, University Medicine Berlin St. Hedwig Hospital, Große Hamburger Str. 5-11 10115 Berlin, Germany Phone: +49-30-2311-2738 Fax: +49-30-2311-2750 www.danielsenkowski.com From jpnv2006 at gmail.com Fri Aug 19 16:01:16 2011 From: jpnv2006 at gmail.com (Juan Pablo Neira) Date: Fri, 19 Aug 2011 16:01:16 +0200 Subject: [FieldTrip] Forward solution using concentric spheres and BEM models In-Reply-To: <71D8F8A56F37A947912FC86D4D9F00F4633F43AE27@XMAIL1.medads.uk-erlangen.de> References: <71D8F8A56F37A947912FC86D4D9F00F4633F43AE27@XMAIL1.medads.uk-erlangen.de> Message-ID: Thank you Margit for your help, I tried your sugestion to create the geometrical description of the brain, skull and scalp but I am still having the same output bnd = 1x1 struct (pnt and tri just of the brain). I should have bnd = 1x3 struct (pnt and tri of the brain, skull and scalp). This is the information in the command line in matlab not downsampling brain not downsampling scalp not downsampling skull using the segmentation approach using the segmented MRI triangulating the boundary of compartment 1 segmentation compartment 1 of 1 completed Can you tell me which version of fieldtrip (fieldtrip-yyyymmdd) are you using so i could compare the code of the function ft_prepare_mesh. Regards, Juan Pablo 2011/8/18 Schönherr, Margit : > Hello, > > I probably cannot help you with all your questions, but currently I'm also dealing with MRI segmentation and volume conductors, so maybe my solutions can be useful for you. > Concerning your point 5 (geometrical description of the brain, skull, and scalp), I have done it the following way: > > cfg             = []; > cfg.method      = 'segmentation'; > cfg.tissue      = [1 2 3]; % scalp = 1; skull = 2; brain = 3; > cfg.numvertices = [2000 1000 800]; > cfg.sourceunits = 'mm'; > cfg.mriunits    = 'mm'; > bnd = ft_prepare_mesh(cfg, mri_segment); > > I think you have to specify a vector 'cfg.numvertices', a number of vertices for each compartment. > By the way, at the beginning of ft_prepare_mesh the cfg-field 'numcompartments' is removed from the cfg structure (at least in my fieldtrip version, which is not the latest). > > Another question that comes to me is, why you need the segmented MRI for a concentric spheres volume conductor? So, regarding point 6, you can build such a vol-structure for example like this: > > vol.unit = 'm'; > vol.o = origin; > vol.r = [R_brain R_skull R_skin]; > vol.c = [1 1/80 1]; > > By this, you can also define the radius as you like. > > Best, > Margit > > > > > ________________________________________ > Von: fieldtrip-bounces at donders.ru.nl [fieldtrip-bounces at donders.ru.nl] im Auftrag von Juan Pablo Neira [jpnv2006 at gmail.com] > Gesendet: Mittwoch, 17. August 2011 14:58 > An: fieldtrip at donders.ru.nl > Betreff: [FieldTrip] Forward solution using concentric spheres and BEM models > > Hello, > > I am working with an individual MRI and EEG data to do the forward > solution with two different volume's model (concentric spheres and > BEM), > so at the end i can compare the results.  But i did not have the > expected results of the concentric spheres model until now. > > This is my script: > > 1.  Read individual MRI > mri=ft_read_mri('data_patient_1\******.hdr'); > > %I got this "Warning: flipping 1st dimension (L-R) to obtain volume in > neurological convention" > > %How can i do a homogenous transformation matrix, using the voxel dimensions > %that are specified in hdr.dime.pixdim? > > 2. Realign to 'head coordinates' > > 3. MRI Reslicing > > 4. MRI segmentation: (brain, skull,scalp) > > 5. Create geometrical description of the brain, skull and scalp (3 compartments) > > cfg                 = []; > cfg.interactive     = 'no';     %segmentation method > cfg.numvertices     = 3000; > cfg.sourceunits     = 'mm'; > cfg.downsample      = 2; > cfg.numcompartments = 3; > cfg.tissue          = {'brain' 'skull' 'scalp'}; > cfg.smooth          = 'yes'; > bnd                 = ft_prepare_mesh(cfg, mri_segment); > > %The output just give me the geometrical description of the brain. > How can i get the geometrical description of > %the skull and scalp also to proceed to create a 3 spheres concentric > sphere model? > > 6.  If i have the 3 geometrical descriptions (brain, skull and scalp). >  Create a 3 concentric spheres model. > > cfg                 = []; > cfg.headshape       = [bnd1 bnd2 bnd3];   % brain- skull - scalp > cfg.conductivity    = [0.33 0.0042 0.331]; > [vol, cfg]          = prepare_concentricspheres(cfg); > > %I got the model but part of the brain is outside of its sphere and > also the skull.  How can I increase the radius so > % the whole brain will be inside its sphere and also the skull?  The > sphere of the scalp fix good. > > I hope someone can help me solve these questions. > > Regards, > > Juan Pablo > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From Margit.Schoenherr at uk-erlangen.de Fri Aug 19 17:23:34 2011 From: Margit.Schoenherr at uk-erlangen.de (=?iso-8859-1?Q?Sch=F6nherr=2C_Margit?=) Date: Fri, 19 Aug 2011 17:23:34 +0200 Subject: [FieldTrip] Forward solution using concentric spheres and BEM models In-Reply-To: References: <71D8F8A56F37A947912FC86D4D9F00F4633F43AE27@XMAIL1.medads.uk-erlangen.de>, Message-ID: <71D8F8A56F37A947912FC86D4D9F00F4633F43AE2A@XMAIL1.medads.uk-erlangen.de> Hello Juan Pablo, I am using fieldtrip-20110603. Now that you say, that your problem persisted, I remember having changed the code of prepare_mesh_segmentation. You find the function in the private directory. In line 25, it sets cfg.tissue = 1 if mri has fields 'brain' or 'scalp'. I think this explains why the segmentation is done for compartment 1 only. So I have simply commented this line, so that cfg.tissue is left unchanged, and the segmentation is done for all compartments. I know that this is bad style, but I wanted to get it working. And since I was relatively new to fieldtrip, I hesitated to ask the discussion list... I hope this helps. And I'm open for better solutions. Best, Margit ________________________________________ Von: fieldtrip-bounces at donders.ru.nl [fieldtrip-bounces at donders.ru.nl] im Auftrag von Juan Pablo Neira [jpnv2006 at gmail.com] Gesendet: Freitag, 19. August 2011 16:01 An: Email discussion list for the FieldTrip project Betreff: Re: [FieldTrip] Forward solution using concentric spheres and BEM models Thank you Margit for your help, I tried your sugestion to create the geometrical description of the brain, skull and scalp but I am still having the same output bnd = 1x1 struct (pnt and tri just of the brain). I should have bnd = 1x3 struct (pnt and tri of the brain, skull and scalp). This is the information in the command line in matlab not downsampling brain not downsampling scalp not downsampling skull using the segmentation approach using the segmented MRI triangulating the boundary of compartment 1 segmentation compartment 1 of 1 completed Can you tell me which version of fieldtrip (fieldtrip-yyyymmdd) are you using so i could compare the code of the function ft_prepare_mesh. Regards, Juan Pablo 2011/8/18 Schönherr, Margit : > Hello, > > I probably cannot help you with all your questions, but currently I'm also dealing with MRI segmentation and volume conductors, so maybe my solutions can be useful for you. > Concerning your point 5 (geometrical description of the brain, skull, and scalp), I have done it the following way: > > cfg = []; > cfg.method = 'segmentation'; > cfg.tissue = [1 2 3]; % scalp = 1; skull = 2; brain = 3; > cfg.numvertices = [2000 1000 800]; > cfg.sourceunits = 'mm'; > cfg.mriunits = 'mm'; > bnd = ft_prepare_mesh(cfg, mri_segment); > > I think you have to specify a vector 'cfg.numvertices', a number of vertices for each compartment. > By the way, at the beginning of ft_prepare_mesh the cfg-field 'numcompartments' is removed from the cfg structure (at least in my fieldtrip version, which is not the latest). > > Another question that comes to me is, why you need the segmented MRI for a concentric spheres volume conductor? So, regarding point 6, you can build such a vol-structure for example like this: > > vol.unit = 'm'; > vol.o = origin; > vol.r = [R_brain R_skull R_skin]; > vol.c = [1 1/80 1]; > > By this, you can also define the radius as you like. > > Best, > Margit > > > > > ________________________________________ > Von: fieldtrip-bounces at donders.ru.nl [fieldtrip-bounces at donders.ru.nl] im Auftrag von Juan Pablo Neira [jpnv2006 at gmail.com] > Gesendet: Mittwoch, 17. August 2011 14:58 > An: fieldtrip at donders.ru.nl > Betreff: [FieldTrip] Forward solution using concentric spheres and BEM models > > Hello, > > I am working with an individual MRI and EEG data to do the forward > solution with two different volume's model (concentric spheres and > BEM), > so at the end i can compare the results. But i did not have the > expected results of the concentric spheres model until now. > > This is my script: > > 1. Read individual MRI > mri=ft_read_mri('data_patient_1\******.hdr'); > > %I got this "Warning: flipping 1st dimension (L-R) to obtain volume in > neurological convention" > > %How can i do a homogenous transformation matrix, using the voxel dimensions > %that are specified in hdr.dime.pixdim? > > 2. Realign to 'head coordinates' > > 3. MRI Reslicing > > 4. MRI segmentation: (brain, skull,scalp) > > 5. Create geometrical description of the brain, skull and scalp (3 compartments) > > cfg = []; > cfg.interactive = 'no'; %segmentation method > cfg.numvertices = 3000; > cfg.sourceunits = 'mm'; > cfg.downsample = 2; > cfg.numcompartments = 3; > cfg.tissue = {'brain' 'skull' 'scalp'}; > cfg.smooth = 'yes'; > bnd = ft_prepare_mesh(cfg, mri_segment); > > %The output just give me the geometrical description of the brain. > How can i get the geometrical description of > %the skull and scalp also to proceed to create a 3 spheres concentric > sphere model? > > 6. If i have the 3 geometrical descriptions (brain, skull and scalp). > Create a 3 concentric spheres model. > > cfg = []; > cfg.headshape = [bnd1 bnd2 bnd3]; % brain- skull - scalp > cfg.conductivity = [0.33 0.0042 0.331]; > [vol, cfg] = prepare_concentricspheres(cfg); > > %I got the model but part of the brain is outside of its sphere and > also the skull. How can I increase the radius so > % the whole brain will be inside its sphere and also the skull? The > sphere of the scalp fix good. > > I hope someone can help me solve these questions. > > Regards, > > Juan Pablo > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From P.Praamstra at neuro.umcn.nl Fri Aug 19 21:04:29 2011 From: P.Praamstra at neuro.umcn.nl (P.Praamstra at neuro.umcn.nl) Date: Fri, 19 Aug 2011 21:04:29 +0200 Subject: [FieldTrip] PhD and post-doctoral position at Donders Centre for Cognitive Neuroimaging Message-ID: Please find attached two adverts for positions at the Donders Centre for Cognitive Neuroimaging, Radboud University Nijmegen, The Netherlands. For both positions we are looking for candidates with MEG or EEG experience. Post-doctoral researcher on project "Movement selection in the parietofrontal cortex: The affordance competition hypothesis". PhD student for project "Feeling the beat: The neurophysiology of cueing in Parkinson's disease". Peter Praamstra, MD, PhD Department of Neurology Radboud University Nijmegen Medical Centre PO Box 9101, 6500 HB Nijmegen The Netherlands Tel: 00-31-24-3668254 Fax: 00-31-24-3541122 Email: p.praamstra at neuro.umcn.nl Het UMC St Radboud staat geregistreerd bij de Kamer van Koophandel in het handelsregister onder nummer 41055629. The Radboud University Nijmegen Medical Centre is listed in the Commercial Register of the Chamber of Commerce under file number 41055629. -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: Advert_Post-doc.doc Type: application/msword Size: 33280 bytes Desc: Advert_Post-doc.doc URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: Advert_AiO.doc Type: application/msword Size: 33280 bytes Desc: Advert_AiO.doc URL: From lmoranr at gmail.com Tue Aug 23 12:02:33 2011 From: lmoranr at gmail.com (=?ISO-8859-1?Q?Luis_Mor=E1n?=) Date: Tue, 23 Aug 2011 12:02:33 +0200 Subject: [FieldTrip] Doubt about de units Message-ID: Dear fieldtrippers, I'm doing a simulation with somato data from MEG. I know that measured fields in data are in* femtoteslas*. Well, the procedure is: - Load data - After setting the configuration structure "cfg", I make a dipolesimulation, and then timelockanalysis, i.e.: *raw1 = dipolesimulation(cfg) avg1=timelockanalysis(cfg, raw1)* What I wonder to know is which are the resultant units of the field * avg1.avg* In the same way, then I perform a reconstruction using beamforming: *sourceAvg1 = sourceanalysis(cfg_beamforming,avg1);* And also, I would like to know the units in which the field * sourceAvg1.avg.pow* is measured Thanks in advance Regards -- Luis Morán Rodríguez -------------- next part -------------- An HTML attachment was scrubbed... URL: From Antony.Passaro at uth.tmc.edu Tue Aug 23 20:50:47 2011 From: Antony.Passaro at uth.tmc.edu (Passaro, Antony D) Date: Tue, 23 Aug 2011 13:50:47 -0500 Subject: [FieldTrip] Source Timecourse In-Reply-To: References: Message-ID: Hi Fieldtrippers, I was looking for a way to extract a time-course from each source estimated by the lcmv beamformer. I searched through the mailing list archives and came across a few emails which talked about the possibility but I was unable to find any examples explaining exactly how this process is done. I am interested in extracting the time-course over the average response rather than the individual trials, if possible. Any help would be greatly appreciated. Thanks, -Tony -------------- next part -------------- An HTML attachment was scrubbed... URL: From Don.Rojas at ucdenver.edu Thu Aug 25 19:43:03 2011 From: Don.Rojas at ucdenver.edu (Rojas, Don) Date: Thu, 25 Aug 2011 11:43:03 -0600 Subject: [FieldTrip] Source Timecourse In-Reply-To: References: Message-ID: <51AF637A-B1EB-4421-84B3-6B0D0C4BBA24@ucdenver.edu> Tony, I don't use the LCMV beamformer, so I can't tell you what the structure of the output is, but what you need to perform a source space projection is the orientation and location information for the sources. In the DICS beamformer output, which I'm familiar with, this information would be in the source structure as follows (assuming your source output structure is called "source"): source.pos % the position info as an n x 3 matrix, where n is the number of sources, in the same units as your volume conductor model source.avg.ori % the orientation as an 1 x n cell array, where n is the number of locations and each cell is a 3 x 1 array of x y z orientation and n refers to the % same source in each array So, in that case, you might find the necessary information for the peak source, in my case, as follows: [~,ind]=max(source.avg.pow(:)); % index of the source with the maximum power in the dics output, let's say ind = 5000 ori=source.avg.ori(5000); % orientation info loc=source.pos(5000,:); % location information Then, if you get this info, you can search the archives for my posts, one of which has a detailed explanation for getting the source waveform when you have a known source location, source orientation and a volume conductor. If you want to get all the waveforms, you can of course iterate the process over the entire voxel volume, but since the leadfields are highly correlated in adjacent source locations, you will see that many of the waveforms will look essentially the same for sources within a few centimeters. Best, Don ----------------------- Don Rojas, Ph.D. Associate Professor of Psychiatry U. of Colorado Denver Anschutz Medical Campus Director, UCD Magnetoencephalography Lab 13001 E. 17th Pl F546 Aurora, CO 80045 USA 303-724-4994 On Aug 23, 2011, at 12:50 PM, Passaro, Antony D wrote: Hi Fieldtrippers, I was looking for a way to extract a time-course from each source estimated by the lcmv beamformer. I searched through the mailing list archives and came across a few emails which talked about the possibility but I was unable to find any examples explaining exactly how this process is done. I am interested in extracting the time-course over the average response rather than the individual trials, if possible. Any help would be greatly appreciated. Thanks, -Tony _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip -------------- next part -------------- An HTML attachment was scrubbed... URL: From ekanal at cmu.edu Fri Aug 26 18:19:45 2011 From: ekanal at cmu.edu (Kanal Eliezer) Date: Fri, 26 Aug 2011 12:19:45 -0400 Subject: [FieldTrip] assessing significance in using ft_timelockanalysis results Message-ID: <8FC0FE65-8E0D-4D7C-9A62-CE1610845C92@cmu.edu> Hello folks - The event related statistics tutorial (http://fieldtrip.fcdonders.nl/tutorial/eventrelatedstatistics) talks about assessing significance parametrically by running t-tests on pooled timelockanalysis data. My question is, does the fact that the averages were created from N trials make a difference? If I'm condition A has twelve averages and condition B has another twelve, and each average contains 70 trials, is there a way to "inform" the statistical test that the power in this dataset is greater than 24? Is this only possible if I run the t-test comparing each set of 840 (70*12) trials? I'm also curious whether this is possible with non-parametric analyses, as well. Thanks - Elli Kanal -------------------- Eliezer Kanal, Ph.D. Postdoctoral Fellow Center for the Neural Basis of Cognition Carnegie Mellon University 4400 Fifth Ave, Suite 110A Pittsburgh PA 15213 P: 412-268-4115 F: 412-268-5060 From e.maris at donders.ru.nl Fri Aug 26 21:00:18 2011 From: e.maris at donders.ru.nl (Eric Maris) Date: Fri, 26 Aug 2011 21:00:18 +0200 Subject: [FieldTrip] assessing significance in using ft_timelockanalysis results In-Reply-To: <8FC0FE65-8E0D-4D7C-9A62-CE1610845C92@cmu.edu> References: <8FC0FE65-8E0D-4D7C-9A62-CE1610845C92@cmu.edu> Message-ID: <020701cc6422$65b10f30$31132d90$@maris@donders.ru.nl> Hi Kanal, > The event related statistics tutorial > (http://fieldtrip.fcdonders.nl/tutorial/eventrelatedstatistics) talks about > assessing significance parametrically by running t-tests on pooled > timelockanalysis data. My question is, does the fact that the averages were > created from N trials make a difference? If I'm condition A has twelve > averages and condition B has another twelve, and each average contains 70 > trials, is there a way to "inform" the statistical test that the power in this > dataset is greater than 24? Is this only possible if I run the t-test comparing > each set of 840 (70*12) trials? > > I'm also curious whether this is possible with non-parametric analyses, as > well. Thanks - In an analysis over subjects (called random-effects analysis in the fMRI literature), "informing" the statistical test about the number of trials per condition only makes sense if this number is different for the two conditions. I propose that you have a look the fMRI papers that deal with the issue of fixed-versus-random effect analyses. The conceptual issues involved are the same in fMRI and electrophysiology. Best, Eric Maris > > Elli Kanal > > > -------------------- > Eliezer Kanal, Ph.D. > Postdoctoral Fellow > Center for the Neural Basis of Cognition > Carnegie Mellon University > 4400 Fifth Ave, Suite 110A > Pittsburgh PA 15213 > P: 412-268-4115 > F: 412-268-5060 > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From tim.curran at Colorado.EDU Sat Aug 27 21:31:25 2011 From: tim.curran at Colorado.EDU (Tim Curran) Date: Sat, 27 Aug 2011 13:31:25 -0600 Subject: [FieldTrip] Fwd: Postdoctoral Position Job Announcement for Bellugi at Salk Institute References: <34FBC147-2620-42D5-ACE8-0D5FD88AC5BD@eng.ucsd.edu> Message-ID: <1686575C-9B5C-4ACB-B752-EF360CF1B8B0@colorado.edu> > > > > -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: Job Ad Final Version_06_14_11.pdf Type: application/pdf Size: 728679 bytes Desc: Job Ad Final Version_06_14_11.pdf URL: From r.oostenveld at donders.ru.nl Mon Aug 29 13:24:21 2011 From: r.oostenveld at donders.ru.nl (Robert Oostenveld) Date: Mon, 29 Aug 2011 13:24:21 +0200 Subject: [FieldTrip] Source Timecourse In-Reply-To: <51AF637A-B1EB-4421-84B3-6B0D0C4BBA24@ucdenver.edu> References: <51AF637A-B1EB-4421-84B3-6B0D0C4BBA24@ucdenver.edu> Message-ID: Dear Tony The lcmv beamformer (by default) estimates the time series of the source, given that the source has an unknown orientation. The result is a 3xNtime matrix as the source activation, which is in source.avg.mom (or in source.trial(i).mom). The three rows of the moment are the strength of the source in the x-, y- and z-direction. Using ft_sourcedescriptives with the option cfg.projectmom=yes you can project the source to its strongest orientation, i.e. the direction that explains most of the source variance. That is equivalent to taking the largest eigenvector of the source timeseries (which is a phrasing that is often used in papers, also on combining fMRI bold timeseries over multiple voxels). You can also do it manually, like this mom3d = randn(3,100); % example source dipole moment [u, s, v] = svd(mom, 'econ'); mom1d = v(1,:); Alternative to taking the largest eigenvector (which has a sign-ambiguity), you can do something like mom3d = randn(3,100); % example source dipole moment [u, s, v] = svd(mom, 'econ'); mom1d = abs(v(1,:)); to take the absolute value, or mom3d = randn(3,100); % example source dipole moment mom1d = sqrt(sum(mom3d.^2,1)); to take the strength over all three directions for each timepoint. best Robert On 25 Aug 2011, at 19:43, Rojas, Don wrote: > Tony, > > I don't use the LCMV beamformer, so I can't tell you what the structure of the output is, but what you need to perform a source space projection is the orientation and location information for the sources. In the DICS beamformer output, which I'm familiar with, this information would be in the source structure as follows (assuming your source output structure is called "source"): > > source.pos % the position info as an n x 3 matrix, where n is the number of sources, in the same units as your volume conductor model > source.avg.ori % the orientation as an 1 x n cell array, where n is the number of locations and each cell is a 3 x 1 array of x y z orientation and n refers to the % same source in each array > > So, in that case, you might find the necessary information for the peak source, in my case, as follows: > > [~,ind]=max(source.avg.pow(:)); % index of the source with the maximum power in the dics output, let's say ind = 5000 > ori=source.avg.ori(5000); % orientation info > loc=source.pos(5000,:); % location information > > Then, if you get this info, you can search the archives for my posts, one of which has a detailed explanation for getting the source waveform when you have a known source location, source orientation and a volume conductor. If you want to get all the waveforms, you can of course iterate the process over the entire voxel volume, but since the leadfields are highly correlated in adjacent source locations, you will see that many of the waveforms will look essentially the same for sources within a few centimeters. > > Best, > > Don > ----------------------- > Don Rojas, Ph.D. > Associate Professor of Psychiatry > U. of Colorado Denver Anschutz Medical Campus > Director, UCD Magnetoencephalography Lab > 13001 E. 17th Pl F546 > Aurora, CO 80045 USA > 303-724-4994 > > On Aug 23, 2011, at 12:50 PM, Passaro, Antony D wrote: > >> Hi Fieldtrippers, >> >> I was looking for a way to extract a time-course from each source estimated by the lcmv beamformer. I searched through the mailing list archives and came across a few emails which talked about the possibility but I was unable to find any examples explaining exactly how this process is done. I am interested in extracting the time-course over the average response rather than the individual trials, if possible. Any help would be greatly appreciated. >> >> Thanks, >> -Tony >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From Antony.Passaro at uth.tmc.edu Mon Aug 29 18:33:55 2011 From: Antony.Passaro at uth.tmc.edu (Passaro, Antony D) Date: Mon, 29 Aug 2011 11:33:55 -0500 Subject: [FieldTrip] Source Timecourse In-Reply-To: References: <51AF637A-B1EB-4421-84B3-6B0D0C4BBA24@ucdenver.edu> Message-ID: Hi Robert and Don, Thank you both for your feedback, I really appreciate it. Using the ft_sourcedescriptives function (and projectmom), I was able to extract the time-course for each source. I am still curious about the output though as it appears to represent the entire duration (including the baseline) of the epoch rather than the specified time-window defined during timelockanalysis (just prior to sourceanalysis). That being the case, does this time-course represent the .pow estimated by the lcmv beamformer or is the nai? Or is it simply a source representation of the gradiometers over the entire epoch? Another issue I came across concerns the source interpolation. It seems once I interpolate the sources using a template mri, only the pow and nai fields are interpolated....is there a way to also interpolate the mom field as well or is it a matter of backtracking to figure out which original source corresponds to each interpolated source? Thank you again for your help, -Tony -----Original Message----- From: fieldtrip-bounces at donders.ru.nl [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Robert Oostenveld Sent: Monday, August 29, 2011 6:24 AM To: Email discussion list for the FieldTrip project Subject: Re: [FieldTrip] Source Timecourse Dear Tony The lcmv beamformer (by default) estimates the time series of the source, given that the source has an unknown orientation. The result is a 3xNtime matrix as the source activation, which is in source.avg.mom (or in source.trial(i).mom). The three rows of the moment are the strength of the source in the x-, y- and z-direction. Using ft_sourcedescriptives with the option cfg.projectmom=yes you can project the source to its strongest orientation, i.e. the direction that explains most of the source variance. That is equivalent to taking the largest eigenvector of the source timeseries (which is a phrasing that is often used in papers, also on combining fMRI bold timeseries over multiple voxels). You can also do it manually, like this mom3d = randn(3,100); % example source dipole moment [u, s, v] = svd(mom, 'econ'); mom1d = v(1,:); Alternative to taking the largest eigenvector (which has a sign-ambiguity), you can do something like mom3d = randn(3,100); % example source dipole moment [u, s, v] = svd(mom, 'econ'); mom1d = abs(v(1,:)); to take the absolute value, or mom3d = randn(3,100); % example source dipole moment mom1d = sqrt(sum(mom3d.^2,1)); to take the strength over all three directions for each timepoint. best Robert On 25 Aug 2011, at 19:43, Rojas, Don wrote: > Tony, > > I don't use the LCMV beamformer, so I can't tell you what the structure of the output is, but what you need to perform a source space projection is the orientation and location information for the sources. In the DICS beamformer output, which I'm familiar with, this information would be in the source structure as follows (assuming your source output structure is called "source"): > > source.pos % the position info as an n x 3 matrix, where n is the number of sources, in the same units as your volume conductor model > source.avg.ori % the orientation as an 1 x n cell array, where n is the number of locations and each cell is a 3 x 1 array of x y z orientation and n refers to the % same source in each array > > So, in that case, you might find the necessary information for the peak source, in my case, as follows: > > [~,ind]=max(source.avg.pow(:)); % index of the source with the maximum power in the dics output, let's say ind = 5000 > ori=source.avg.ori(5000); % orientation info > loc=source.pos(5000,:); % location information > > Then, if you get this info, you can search the archives for my posts, one of which has a detailed explanation for getting the source waveform when you have a known source location, source orientation and a volume conductor. If you want to get all the waveforms, you can of course iterate the process over the entire voxel volume, but since the leadfields are highly correlated in adjacent source locations, you will see that many of the waveforms will look essentially the same for sources within a few centimeters. > > Best, > > Don > ----------------------- > Don Rojas, Ph.D. > Associate Professor of Psychiatry > U. of Colorado Denver Anschutz Medical Campus Director, UCD > Magnetoencephalography Lab > 13001 E. 17th Pl F546 > Aurora, CO 80045 USA > 303-724-4994 > > On Aug 23, 2011, at 12:50 PM, Passaro, Antony D wrote: > >> Hi Fieldtrippers, >> >> I was looking for a way to extract a time-course from each source estimated by the lcmv beamformer. I searched through the mailing list archives and came across a few emails which talked about the possibility but I was unable to find any examples explaining exactly how this process is done. I am interested in extracting the time-course over the average response rather than the individual trials, if possible. Any help would be greatly appreciated. >> >> Thanks, >> -Tony >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From irina.simanova at mpi.nl Tue Aug 30 08:03:29 2011 From: irina.simanova at mpi.nl (Irina Simanova) Date: Tue, 30 Aug 2011 08:03:29 +0200 Subject: [FieldTrip] ICA on MEG data Message-ID: <729C68C8-BB67-44F7-B439-A5B9C90BBB9D@mpi.nl> Dear FieldTripers, I would like to identify ocular artifacts in an MEG dataset using the ICA. The data consist of 180 trials of variable duration ( the signal from stimulus onset to subject's response). The trials' duration vary from 400 ms to 3000 ms. The data is sampled at 1.2 kHz, and has 275 channels. I am new to ICA, and I need a piece of advice. What would be a reasonable number of output independent components? Should I do the PCA as a pre-processing stage? How do i find an optimal number of principal components to extract? I would appreciate your help, Kind regards, Irina Simanova PhD student Neurobiology of Language Group Max-Planck Institute for Psycholinguistics Wundtlaan 1 6525 XD Nijmegen The Netherlands e-mail: irina.simanova at mpi.nl phone: +31 24 3521541 From r.oostenveld at donders.ru.nl Tue Aug 30 09:08:19 2011 From: r.oostenveld at donders.ru.nl (Robert Oostenveld) Date: Tue, 30 Aug 2011 09:08:19 +0200 Subject: [FieldTrip] Source Timecourse In-Reply-To: References: <51AF637A-B1EB-4421-84B3-6B0D0C4BBA24@ucdenver.edu> Message-ID: <9370AD38-AA89-4038-957B-28525CAA0A37@donders.ru.nl> Hi Tony On 29 Aug 2011, at 18:33, Passaro, Antony D wrote: > Thank you both for your feedback, I really appreciate it. Using the ft_sourcedescriptives function (and projectmom), I was able to extract the time-course for each source. I am still curious about the output though as it appears to represent the entire duration (including the baseline) of the epoch rather than the specified time-window defined during timelockanalysis (just prior to sourceanalysis). Default behaviour is to construct the covariance and average for the whole timewindow, indeed including the baseline period. Depending on the experimental data, the LCMV filter can be optimized for a specific timewindow by computing the covariance only for that timewindow. Subsequently the filter can be applied to the whole timewindow (including baseline) or only to the timewindow of interest. > That being the case, does this time-course represent the .pow estimated by the lcmv beamformer or is the nai? source.avg.pow is the power estimated based on the data covariance window, not on the longer time window that you might pass through the filter. If you computed the covariance for a smaller timewindow (using ft_timelockanalysis), then only that is used for the power estimate. The same applies to the nai. > Or is it simply a source representation of the gradiometers over the entire epoch? Another issue I came across concerns the source interpolation. It seems once I interpolate the sources using a template mri, only the pow and nai fields are interpolated....is there a way to also interpolate the mom field as well or is it a matter of backtracking to figure out which original source corresponds to each interpolated source? The mom timecourse is indeed not interpolated, as that used to exceed the memory on many computers: the interpolated mom would consist of a full 3D volume at each timepoint of your ERF. We are in the process of making some improvements to the source data structure, i.e. to its representation. The planning is mostly done, but implementation wise I don't know how far the code already is. I think you could work with the following to interpolate it source.avg.momint = nan(length(source.pos,1), length(source.time)); for i=1:length(source.inside) indx = source.inside(i); source.avg.momint(indx,:) = source.avg.mom{indx}; % the grid locations outside the brain keep their NaN end and then specify cfg.parameter = 'momint' in ft_sourceinterpolate. best Robert From stan.vanpelt at fcdonders.ru.nl Tue Aug 30 09:14:31 2011 From: stan.vanpelt at fcdonders.ru.nl (Stan van Pelt) Date: Tue, 30 Aug 2011 09:14:31 +0200 (CEST) Subject: [FieldTrip] ICA on MEG data In-Reply-To: <729C68C8-BB67-44F7-B439-A5B9C90BBB9D@mpi.nl> References: <729C68C8-BB67-44F7-B439-A5B9C90BBB9D@mpi.nl> Message-ID: <0db501cc66e4$76b60a00$64221e00$@vanpelt@fcdonders.ru.nl> Dear Irina, There's a page on the Fieldtrip Wiki about using ICA to remove EOG-artifacts, see http://fieldtrip.fcdonders.nl/example/use_independent_component_analysis_i ca_to_remove_eog_artifacts. The common approach is to first define your trials, and then run ICA on them. You will get as many output components as you have channels, i.e. 275 in your case. The EOG artifacts usually pop up in 1 or 2 components within the first 20 largest components. Best regards, Stan Stan van Pelt, PhD Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen Kapittelweg 29, 6525 EN Nijmegen, Netherlands E-mail: stan.vanpelt at donders.ru.nl Website: www.ru.nl/donders/ Phone: (+31) (0)24 36 68495 Fax: (+31) (0)24 36 10989 -----Original Message----- From: fieldtrip-bounces at donders.ru.nl [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Irina Simanova Sent: Tuesday, August 30, 2011 8:03 AM To: Email discussion list for the FieldTrip project Subject: [FieldTrip] ICA on MEG data Dear FieldTripers, I would like to identify ocular artifacts in an MEG dataset using the ICA. The data consist of 180 trials of variable duration ( the signal from stimulus onset to subject's response). The trials' duration vary from 400 ms to 3000 ms. The data is sampled at 1.2 kHz, and has 275 channels. I am new to ICA, and I need a piece of advice. What would be a reasonable number of output independent components? Should I do the PCA as a pre-processing stage? How do i find an optimal number of principal components to extract? I would appreciate your help, Kind regards, Irina Simanova PhD student Neurobiology of Language Group Max-Planck Institute for Psycholinguistics Wundtlaan 1 6525 XD Nijmegen The Netherlands e-mail: irina.simanova at mpi.nl phone: +31 24 3521541 _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From irina.simanova at mpi.nl Tue Aug 30 09:56:12 2011 From: irina.simanova at mpi.nl (Irina Simanova) Date: Tue, 30 Aug 2011 09:56:12 +0200 Subject: [FieldTrip] ICA on MEG data In-Reply-To: <0db501cc66e4$76b60a00$64221e00$@vanpelt@fcdonders.ru.nl> References: <729C68C8-BB67-44F7-B439-A5B9C90BBB9D@mpi.nl> <0db501cc66e4$76b60a00$64221e00$@vanpelt@fcdonders.ru.nl> Message-ID: <79A92C58-4FA2-4ADD-B6D6-975A4C4B9F08@mpi.nl> Dear Stan, Thank you for your reply, As far as I understand from the tutorials about ICA by Scott Makeig http://sccn.ucsd.edu/~scott/tutorial/icafaq.html , the number of time points should be considerably larger than the number of scalp sensors. How critical is this point? Besides, the sensors are very close to each other, and the registered activity in adjacent sensors is highly correlated, so the components would be similar, right? Isn't it reasonable to reduce the number of output components then? Thank you in advance, Irina On Aug 30, 2011, at 9:14 AM 8/30/11, Stan van Pelt wrote: > Dear Irina, > > There's a page on the Fieldtrip Wiki about using ICA to remove > EOG-artifacts, see > http://fieldtrip.fcdonders.nl/example/use_independent_component_analysis_i > ca_to_remove_eog_artifacts. > > The common approach is to first define your trials, and then run ICA > on > them. > > You will get as many output components as you have channels, i.e. > 275 in > your case. The EOG artifacts usually pop up in 1 or 2 components > within > the first 20 largest components. > > Best regards, > Stan > > Stan van Pelt, PhD > Donders Institute for Brain, Cognition and Behaviour, Radboud > University > Nijmegen > Kapittelweg 29, 6525 EN Nijmegen, Netherlands > E-mail: stan.vanpelt at donders.ru.nl > Website: www.ru.nl/donders/ > Phone: (+31) (0)24 36 68495 > Fax: (+31) (0)24 36 10989 > > -----Original Message----- > From: fieldtrip-bounces at donders.ru.nl > [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Irina Simanova > Sent: Tuesday, August 30, 2011 8:03 AM > To: Email discussion list for the FieldTrip project > Subject: [FieldTrip] ICA on MEG data > > Dear FieldTripers, > > I would like to identify ocular artifacts in an MEG dataset using the > ICA. > > The data consist of 180 trials of variable duration ( the signal from > stimulus onset to subject's response). The trials' duration vary from > 400 ms to 3000 ms. The data is sampled at 1.2 kHz, and has 275 > channels. I am new to ICA, and I need a piece of advice. What would be > a reasonable number of output independent components? > Should I do the PCA as a pre-processing stage? How do i find an > optimal number of principal components to extract? > > I would appreciate your help, > > Kind regards, > > Irina Simanova > PhD student > Neurobiology of Language Group > Max-Planck Institute for Psycholinguistics > Wundtlaan 1 > 6525 XD Nijmegen > The Netherlands > e-mail: irina.simanova at mpi.nl > phone: +31 24 3521541 > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From stan.vanpelt at fcdonders.ru.nl Tue Aug 30 10:15:07 2011 From: stan.vanpelt at fcdonders.ru.nl (Stan van Pelt) Date: Tue, 30 Aug 2011 10:15:07 +0200 (CEST) Subject: [FieldTrip] ICA on MEG data In-Reply-To: <79A92C58-4FA2-4ADD-B6D6-975A4C4B9F08@mpi.nl> References: <729C68C8-BB67-44F7-B439-A5B9C90BBB9D@mpi.nl> <0db501cc66e4$76b60a00$64221e00$@vanpelt@fcdonders.ru.nl> <79A92C58-4FA2-4ADD-B6D6-975A4C4B9F08@mpi.nl> Message-ID: <0e0601cc66ec$ee2048f0$ca60dad0$@vanpelt@fcdonders.ru.nl> Dear Irina, The number of time points should indeed be much larger for a good ICA estimate. However, that seems to be the case with your data, so I think there's no problem there. As far as I know, with ICA in fieldtrip you always get the same amount of output components as input channels. Activity in adjacent MEG channels will be somewhat correlated indeed, but not nearly as much as with EEG. Moreover, the origin of your artifacts (EOG, maybe also ECG) will create a distinct topographic distribution, and the components will pop out clearly from your ft_componentanalysis, because it explains large amounts of variance in the data. Just give it a try, I'd say. You should get similar results as in the Wiki example. Best, Stan -----Original Message----- From: fieldtrip-bounces at donders.ru.nl [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Irina Simanova Sent: Tuesday, August 30, 2011 9:56 AM To: Email discussion list for the FieldTrip project Subject: Re: [FieldTrip] ICA on MEG data Dear Stan, Thank you for your reply, As far as I understand from the tutorials about ICA by Scott Makeig http://sccn.ucsd.edu/~scott/tutorial/icafaq.html , the number of time points should be considerably larger than the number of scalp sensors. How critical is this point? Besides, the sensors are very close to each other, and the registered activity in adjacent sensors is highly correlated, so the components would be similar, right? Isn't it reasonable to reduce the number of output components then? Thank you in advance, Irina On Aug 30, 2011, at 9:14 AM 8/30/11, Stan van Pelt wrote: > Dear Irina, > > There's a page on the Fieldtrip Wiki about using ICA to remove > EOG-artifacts, see > http://fieldtrip.fcdonders.nl/example/use_independent_component_analysis_i > ca_to_remove_eog_artifacts. > > The common approach is to first define your trials, and then run ICA > on > them. > > You will get as many output components as you have channels, i.e. > 275 in > your case. The EOG artifacts usually pop up in 1 or 2 components > within > the first 20 largest components. > > Best regards, > Stan > > Stan van Pelt, PhD > Donders Institute for Brain, Cognition and Behaviour, Radboud > University > Nijmegen > Kapittelweg 29, 6525 EN Nijmegen, Netherlands > E-mail: stan.vanpelt at donders.ru.nl > Website: www.ru.nl/donders/ > Phone: (+31) (0)24 36 68495 > Fax: (+31) (0)24 36 10989 > > -----Original Message----- > From: fieldtrip-bounces at donders.ru.nl > [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Irina Simanova > Sent: Tuesday, August 30, 2011 8:03 AM > To: Email discussion list for the FieldTrip project > Subject: [FieldTrip] ICA on MEG data > > Dear FieldTripers, > > I would like to identify ocular artifacts in an MEG dataset using the > ICA. > > The data consist of 180 trials of variable duration ( the signal from > stimulus onset to subject's response). The trials' duration vary from > 400 ms to 3000 ms. The data is sampled at 1.2 kHz, and has 275 > channels. I am new to ICA, and I need a piece of advice. What would be > a reasonable number of output independent components? > Should I do the PCA as a pre-processing stage? How do i find an > optimal number of principal components to extract? > > I would appreciate your help, > > Kind regards, > > Irina Simanova > PhD student > Neurobiology of Language Group > Max-Planck Institute for Psycholinguistics > Wundtlaan 1 > 6525 XD Nijmegen > The Netherlands > e-mail: irina.simanova at mpi.nl > phone: +31 24 3521541 > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From k.muesch at uke.uni-hamburg.de Tue Aug 30 10:44:42 2011 From: k.muesch at uke.uni-hamburg.de (=?iso-8859-1?Q?Kathrin_M=FCsch?=) Date: Tue, 30 Aug 2011 10:44:42 +0200 Subject: [FieldTrip] ICA on MEG data In-Reply-To: <0e0601cc66ec$ee2048f0$ca60dad0$@vanpelt@fcdonders.ru.nl> References: <729C68C8-BB67-44F7-B439-A5B9C90BBB9D@mpi.nl> <0db501cc66e4$76b60a00$64221e00$@vanpelt@fcdonders.ru.nl> <79A92C58-4FA2-4ADD-B6D6-975A4C4B9F08@mpi.nl> <0e0601cc66ec$ee2048f0$ca60dad0$@vanpelt@fcdonders.ru.nl> Message-ID: Dear Irina, I reduced the number of components to 64 by running a PCA beforehand (cfg.runica.pca = 64) on my MEG data. Depending on how much the subjects moved their eyes, blinks and saccades popped up in earlier or later components. I had feasible results with this approach both for EOG and ECG. Good luck, Kathrin _____________________________________ Kathrin Müsch Dept. of Neurophysiology and Pathophysiology University Medical Center Hamburg-Eppendorf Martinistr. 52 20246 Hamburg Germany Phone: +49-40-7410-54680 Fax: +49-40-7410-57752 E-Mail: k.muesch at uke.uni-hamburg.de _____________________________________ Am 30.08.2011 um 10:15 schrieb Stan van Pelt: > Dear Irina, > > The number of time points should indeed be much larger for a good ICA > estimate. However, that seems to be the case with your data, so I think > there's no problem there. > > As far as I know, with ICA in fieldtrip you always get the same amount of > output components as input channels. Activity in adjacent MEG channels > will be somewhat correlated indeed, but not nearly as much as with EEG. > Moreover, the origin of your artifacts (EOG, maybe also ECG) will create a > distinct topographic distribution, and the components will pop out clearly > from your ft_componentanalysis, because it explains large amounts of > variance in the data. > > Just give it a try, I'd say. You should get similar results as in the Wiki > example. > > Best, > Stan > > > -----Original Message----- > From: fieldtrip-bounces at donders.ru.nl > [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Irina Simanova > Sent: Tuesday, August 30, 2011 9:56 AM > To: Email discussion list for the FieldTrip project > Subject: Re: [FieldTrip] ICA on MEG data > > Dear Stan, > > Thank you for your reply, > > As far as I understand from the tutorials about ICA by Scott Makeig > http://sccn.ucsd.edu/~scott/tutorial/icafaq.html > , the number of time points should be considerably larger than the > number of scalp sensors. How critical is this point? > > Besides, the sensors are very close to each other, and the registered > activity in adjacent sensors is highly correlated, so the components > would be similar, right? Isn't it reasonable to reduce the number of > output components then? > > Thank you in advance, > Irina > > > > On Aug 30, 2011, at 9:14 AM 8/30/11, Stan van Pelt wrote: > >> Dear Irina, >> >> There's a page on the Fieldtrip Wiki about using ICA to remove >> EOG-artifacts, see >> > http://fieldtrip.fcdonders.nl/example/use_independent_component_analysis_i >> ca_to_remove_eog_artifacts. >> >> The common approach is to first define your trials, and then run ICA >> on >> them. >> >> You will get as many output components as you have channels, i.e. >> 275 in >> your case. The EOG artifacts usually pop up in 1 or 2 components >> within >> the first 20 largest components. >> >> Best regards, >> Stan >> >> Stan van Pelt, PhD >> Donders Institute for Brain, Cognition and Behaviour, Radboud >> University >> Nijmegen >> Kapittelweg 29, 6525 EN Nijmegen, Netherlands >> E-mail: stan.vanpelt at donders.ru.nl >> Website: www.ru.nl/donders/ >> Phone: (+31) (0)24 36 68495 >> Fax: (+31) (0)24 36 10989 >> >> -----Original Message----- >> From: fieldtrip-bounces at donders.ru.nl >> [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Irina Simanova >> Sent: Tuesday, August 30, 2011 8:03 AM >> To: Email discussion list for the FieldTrip project >> Subject: [FieldTrip] ICA on MEG data >> >> Dear FieldTripers, >> >> I would like to identify ocular artifacts in an MEG dataset using the >> ICA. >> >> The data consist of 180 trials of variable duration ( the signal from >> stimulus onset to subject's response). The trials' duration vary from >> 400 ms to 3000 ms. The data is sampled at 1.2 kHz, and has 275 >> channels. I am new to ICA, and I need a piece of advice. What would be >> a reasonable number of output independent components? >> Should I do the PCA as a pre-processing stage? How do i find an >> optimal number of principal components to extract? >> >> I would appreciate your help, >> >> Kind regards, >> >> Irina Simanova >> PhD student >> Neurobiology of Language Group >> Max-Planck Institute for Psycholinguistics >> Wundtlaan 1 >> 6525 XD Nijmegen >> The Netherlands >> e-mail: irina.simanova at mpi.nl >> phone: +31 24 3521541 >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From irina.simanova at mpi.nl Tue Aug 30 14:22:31 2011 From: irina.simanova at mpi.nl (Irina Simanova) Date: Tue, 30 Aug 2011 14:22:31 +0200 Subject: [FieldTrip] Call for Papers :: NIPS 2011 Workshop on Interpretable Decoding of Higher Cognitive States from Neural Data Message-ID: <2E7CF2CC-C7AF-4937-935E-D0337E8445C8@mpi.nl> Interpretable Decoding of Higher Cognitive States from Neural Data NIPS 2011 Workshop, Dec 16 or 17, 2011, Granada, Spain (Please feel free to distribute the CFP to all the interested persons and groups.) Overview Over recent years, machine learning methods have become a crucial analytical tool in cognitive neuroscience (see reviews by Formisano et al., 2008; Pereira et al., 2009). Decoding techniques have dramatically increased the sensitivity of experiments, and so also the subtlety of cognitive questions that can be asked. At the same time the mental phenomena being studied are moving beyond lower-level perceptual and motor processes which are directly grounded in external measurable realities. Decoding higher cognition and interpreting the learned behaviour of the classifiers used pose unique challenges, as these psychological states are complex, fast-changing and often ill-defined. Contemporary machine learning methods deal well with the small numbers of cases, and high numbers of co-linear dimensions typical of neural data, and are generally optimized to maximize classification performance, rather than to enable meaningful interpretation of the features they learn from. And indeed recent work has succeeded to decode psychological phenomena including visual object recognition (e.g. Kriegeskorte et al., 2008; Connolly et al., 2011), perceptual interpretation of sounds (Staeren et al., 2009), lexical semantics (Mitchell et al., 2008; Siminova et al., 2010; Devereux et al., 2010; Murphy et al., 2011), decision making during game playing (Xiang et al., 2009) and the process of mental arithmetic (Anderson et al., 2008). But for the cognitive scientists who use these methods, the primary question is often not "how much" but rather "how" and "why" the patterns of neural activity identified by a machine learning algorithm encode particular cognitive processes. The aim of this workshop is therefore to 1) discuss the achievements and problems of the decoding of high-level cognitive states, and 2) explore the use of machine learning methodologies and other computational models that enable such cognitive interpretation of neural recordings of different modalities. Advances in this field require close collaboration between machine learning experts, neuroscientists and cognitive scientists. Thus, this workshop is highly interdisciplinary and will aim to attract submissions also from outside the existing NIPS community. By stimulating discussions among experts in the different fields, the workshop seeks to generate novel insights and new directions for research. Topics of interest The field requires techniques that are capable of taking advantage of spatially distributed patterns in the brain, that are separated in space but coordinated in their activity. Methods should also be sensitive to the fine-grained temporal patterns of multiple processes - which may proceed in a serial fashion, overlapping or in parallel with each-other, or in multiple passes with bidirectional information flows. Different recording modalities have distinctive advantages: fMRI provides very fine millimetre-level localisation in the brain but poor temporal resolution, while EEG and MEG have millisecond temporal resolution at the cost of spatial resolution. Ideally machine learning methods would be able to meaningfully combine complementary information from these different neuroimaging techniques (see e.g. De Martino et al., 2010). Moreover, as the processes underlying higher cognition are so complex, methods should be able to disentangle even tightly linked and confounded subprocesses. Finally, general use algorithms that could induce latent dimensions from neural data, and so reveal the "hidden" psychological states, would be a dramatic advance on current hypothesis-driven analytical paradigms. Originality of approach is encouraged and submissions on any related methodological approach are welcomed, such as: - Interpreting spatial and temporal location of selected features and their weights - Discovering "hidden" or "latent" cognitive representations - Disentangling confounded processes and representations - Comparing or combining data from recording modalities (e.g. fMRI, EEG, structural MRI, DTI, MEG, NIRS, EcOG, single cell recordings) - Fuzzy and partial classifications - Unaligned or incommensurate feature spaces and data representation As noted above, the complexity of higher cognition poses challenges. To take language comprehension as an example, speech is received at 3-4 words; acoustic, semantic and syntactic processing can occur in parallel; and the form of underlying representations (sentence structures, conceptual descriptions) remains controversial. We welcome submissions dealing with any high-level cognitive functions that exhibit similar complexity, for instance: - Knowledge representation and concepts - Language and communication - Understanding visual and auditory experience - Memory and learning - Reasoning and problem solving - Decision making and executive control Submissions Authors are invited to submit full papers on original, unpublished work in the topic area of this workshop via the NIPS 2011 submission site at https://cmt.research.microsoft.com/NIPS2011/Default.aspx. Submissions should be formatted using the NIPS 2011 stylefiles, with blind review and not exceeding 8 pages plus an extra page for references. Author and submission information can be found at http://nips.cc/PaperInformation/AuthorSubmissionInstructions . The stylefiles are available at http://nips.cc/PaperInformation/StyleFiles . Each submission will be reviewed at least by two members of the programme committee. Accepted papers will be published in the workshop proceedings. Dual submissions to the main NIPS 2011 conference and this workshop are allowed; if you submit to the main session, indicate this when you submit to the workshop. If your paper is accepted for the main session, you should withdraw your paper from the workshop upon notification by the main session. Important Dates - Aug 30, 2011: Call for papers - Sep 23, 2011: Deadline for submission of workshop papers - Oct 15, 2011: Notification of acceptance - Oct 31, 2011: Camera-ready papers due - Dec 16 or 17, 2011: Workshop date Links - NIPS 2011 website: http://nips.cc/Conferences/2011/ - Workshop website: https://sites.google.com/site/decodehighcogstate - Call for Papers: https://sites.google.com/site/decodehighcogstate/cfp/ Kind regards, The Workshop Organizers, Kai-min Kevin Chang, Anna Korhonen, Irina Simanova, Brian Murphy -------------- next part -------------- An HTML attachment was scrubbed... URL: From anna.lambrechts at gmail.com Tue Aug 30 17:46:52 2011 From: anna.lambrechts at gmail.com (Anna Lambrechts) Date: Tue, 30 Aug 2011 17:46:52 +0200 Subject: [FieldTrip] Issues with neighbours in ft_timelockstatistics Message-ID: Dear fieldtrip community, Until about two weeks ago I was using a batch to process group comparison statistics on MEG data using ft_timelockstatistics. A week later, the same batch was not working anymore, even on the same data. The error I get is the following: *??? Improper index matrix reference. Error in ==> clusterstat>makechanneighbstructmat at 520 [seld] = match_str(cfg.channel, cfg.neighbours(chan).label); Error in ==> clusterstat at 60 channeighbstructmat = makechanneighbstructmat(cfg); Error in ==> statistics_montecarlo at 320 [stat, cfg] = clusterstat(cfg, statrand, statobs,'issource',issource); Error in ==> statistics_wrapper at 290 [stat, cfg] = statmethod(cfg, dat, design, 'issource',issource); Error in ==> ft_timelockstatistics at 124 [stat, cfg] = statistics_wrapper(cfg, varargin{:}); Error in ==> myfunction_stats at 59 gpstat = ft_timelockstatistics(cfg, data1, data2); *It seems the error concerns the 'neighbours.mat' file reading, even though I am not sure about it. When I delete the two last lines in this file (to match the size of matrix wanted) the error disappears, but the processing is incomplete and false (or at least it does not match the results I had before). Has anyone encounter the same issue? Did you do any update recently that might explain this problem, and do you have any idea how to remedy to it? Best wishes, Anna. -------------- next part -------------- An HTML attachment was scrubbed... URL: From ekanal at cmu.edu Tue Aug 30 17:58:34 2011 From: ekanal at cmu.edu (Kanal Eliezer) Date: Tue, 30 Aug 2011 11:58:34 -0400 Subject: [FieldTrip] assessing significance in using ft_timelockanalysis results In-Reply-To: <2109_1314385277_p7QJ1Gjs003376_020701cc6422$65b10f30$31132d90$@maris@donders.ru.nl> References: <8FC0FE65-8E0D-4D7C-9A62-CE1610845C92@cmu.edu> <2109_1314385277_p7QJ1Gjs003376_020701cc6422$65b10f30$31132d90$@maris@donders.ru.nl> Message-ID: <8CC6BA25-9A03-4630-B940-486BE2B9EE92@cmu.edu> Hello Eric - Thanks for the response. It looks like the fixed vs random analysis is exactly what I'm referring to. From what I understood, it looks like the difference really only shows up in the variance of the resultant distribution; with a random, the variance also takes into account the betwee-subjects variance. Is there a way to specify whether I want to do a fixed or random effects analysis in FieldTrip when I'm running ft_timelockgrandaverage or ft_freqgrandaverage? Thanks! Elli p.s. - In case anyone else is trying to figure this out, chapter 12 of Friston's book "Statistical Parametric Mapping" does an excellent job explaining the difference between fixed and random analyses, as well as how to implement it algorithmically. On Aug 26, 2011, at 3:00 PM, Eric Maris wrote: > Hi Kanal, > >> The event related statistics tutorial >> (http://fieldtrip.fcdonders.nl/tutorial/eventrelatedstatistics) talks > about >> assessing significance parametrically by running t-tests on pooled >> timelockanalysis data. My question is, does the fact that the averages > were >> created from N trials make a difference? If I'm condition A has twelve >> averages and condition B has another twelve, and each average contains 70 >> trials, is there a way to "inform" the statistical test that the power in > this >> dataset is greater than 24? Is this only possible if I run the t-test > comparing >> each set of 840 (70*12) trials? >> >> I'm also curious whether this is possible with non-parametric analyses, as >> well. Thanks - > > In an analysis over subjects (called random-effects analysis in the fMRI > literature), "informing" the statistical test about the number of trials per > condition only makes sense if this number is different for the two > conditions. I propose that you have a look the fMRI papers that deal with > the issue of fixed-versus-random effect analyses. The conceptual issues > involved are the same in fMRI and electrophysiology. > > > Best, > > Eric Maris > > > > >> >> Elli Kanal >> >> >> -------------------- >> Eliezer Kanal, Ph.D. >> Postdoctoral Fellow >> Center for the Neural Basis of Cognition >> Carnegie Mellon University >> 4400 Fifth Ave, Suite 110A >> Pittsburgh PA 15213 >> P: 412-268-4115 >> F: 412-268-5060 >> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From e.maris at donders.ru.nl Tue Aug 30 21:13:42 2011 From: e.maris at donders.ru.nl (Eric Maris) Date: Tue, 30 Aug 2011 21:13:42 +0200 Subject: [FieldTrip] assessing significance in using ft_timelockanalysis results Message-ID: <00d201cc6748$f18e8fe0$d4abafa0$@maris@donders.ru.nl> Hi Elli, > Thanks for the response. It looks like the fixed vs random analysis is exactly > what I'm referring to. From what I understood, it looks like the difference > really only shows up in the variance of the resultant distribution; with a > random, the variance also takes into account the betwee-subjects variance. > Is there a way to specify whether I want to do a fixed or random effects > analysis in FieldTrip when I'm running ft_timelockgrandaverage or > ft_freqgrandaverage? Thanks! I don't get this. Ft_timelockgrandaverage and ft_freqgrandaverage will only be called when a random effects analysis is performed. Best, Eric > > Elli > > p.s. - In case anyone else is trying to figure this out, chapter 12 of Friston's > book "Statistical Parametric Mapping" does an excellent job explaining the > difference between fixed and random analyses, as well as how to implement > it algorithmically. > > > > On Aug 26, 2011, at 3:00 PM, Eric Maris wrote: > > > Hi Kanal, > > > >> The event related statistics tutorial > >> (http://fieldtrip.fcdonders.nl/tutorial/eventrelatedstatistics) talks > > about > >> assessing significance parametrically by running t-tests on pooled > >> timelockanalysis data. My question is, does the fact that the averages > > were > >> created from N trials make a difference? If I'm condition A has twelve > >> averages and condition B has another twelve, and each average contains > 70 > >> trials, is there a way to "inform" the statistical test that the power in > > this > >> dataset is greater than 24? Is this only possible if I run the t-test > > comparing > >> each set of 840 (70*12) trials? > >> > >> I'm also curious whether this is possible with non-parametric analyses, as > >> well. Thanks - > > > > In an analysis over subjects (called random-effects analysis in the fMRI > > literature), "informing" the statistical test about the number of trials per > > condition only makes sense if this number is different for the two > > conditions. I propose that you have a look the fMRI papers that deal with > > the issue of fixed-versus-random effect analyses. The conceptual issues > > involved are the same in fMRI and electrophysiology. > > > > > > Best, > > > > Eric Maris > > > > > > > > > >> > >> Elli Kanal > >> > >> > >> -------------------- > >> Eliezer Kanal, Ph.D. > >> Postdoctoral Fellow > >> Center for the Neural Basis of Cognition > >> Carnegie Mellon University > >> 4400 Fifth Ave, Suite 110A > >> Pittsburgh PA 15213 > >> P: 412-268-4115 > >> F: 412-268-5060 > >> > >> > >> _______________________________________________ > >> fieldtrip mailing list > >> fieldtrip at donders.ru.nl > >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > _______________________________________________ > > fieldtrip mailing list > > fieldtrip at donders.ru.nl > > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From ekanal at cmu.edu Tue Aug 30 21:46:24 2011 From: ekanal at cmu.edu (Kanal Eliezer) Date: Tue, 30 Aug 2011 15:46:24 -0400 Subject: [FieldTrip] assessing significance in using ft_timelockanalysis results In-Reply-To: <29602_1314731679_p7UJEcWB023970_00d201cc6748$f18e8fe0$d4abafa0$@maris@donders.ru.nl> References: <29602_1314731679_p7UJEcWB023970_00d201cc6748$f18e8fe0$d4abafa0$@maris@donders.ru.nl> Message-ID: <4D99A829-9BB1-4395-9997-45568224DAEC@cmu.edu> Maybe I had been using it incorrectly, then. In my study, I have numerous datafiles for each subject (each block gets split into it's own .fif file, due to file size), and I use those functions to combine the preprocessed and timelocked datafiles. Should I be doing this differently? Elli On Aug 30, 2011, at 3:13 PM, Eric Maris wrote: > Hi Elli, > > > >> Thanks for the response. It looks like the fixed vs random analysis is > exactly >> what I'm referring to. From what I understood, it looks like the > difference >> really only shows up in the variance of the resultant distribution; with a >> random, the variance also takes into account the betwee-subjects variance. >> Is there a way to specify whether I want to do a fixed or random effects >> analysis in FieldTrip when I'm running ft_timelockgrandaverage or >> ft_freqgrandaverage? Thanks! > > > I don't get this. Ft_timelockgrandaverage and ft_freqgrandaverage will only > be called when a random effects analysis is performed. > > > Best, > > Eric > > >> >> Elli >> >> p.s. - In case anyone else is trying to figure this out, chapter 12 of > Friston's >> book "Statistical Parametric Mapping" does an excellent job explaining the >> difference between fixed and random analyses, as well as how to implement >> it algorithmically. >> >> >> >> On Aug 26, 2011, at 3:00 PM, Eric Maris wrote: >> >>> Hi Kanal, >>> >>>> The event related statistics tutorial >>>> (http://fieldtrip.fcdonders.nl/tutorial/eventrelatedstatistics) talks >>> about >>>> assessing significance parametrically by running t-tests on pooled >>>> timelockanalysis data. My question is, does the fact that the averages >>> were >>>> created from N trials make a difference? If I'm condition A has twelve >>>> averages and condition B has another twelve, and each average contains >> 70 >>>> trials, is there a way to "inform" the statistical test that the power > in >>> this >>>> dataset is greater than 24? Is this only possible if I run the t-test >>> comparing >>>> each set of 840 (70*12) trials? >>>> >>>> I'm also curious whether this is possible with non-parametric analyses, > as >>>> well. Thanks - >>> >>> In an analysis over subjects (called random-effects analysis in the fMRI >>> literature), "informing" the statistical test about the number of trials > per >>> condition only makes sense if this number is different for the two >>> conditions. I propose that you have a look the fMRI papers that deal > with >>> the issue of fixed-versus-random effect analyses. The conceptual issues >>> involved are the same in fMRI and electrophysiology. >>> >>> >>> Best, >>> >>> Eric Maris >>> >>> >>> >>> >>>> >>>> Elli Kanal >>>> >>>> >>>> -------------------- >>>> Eliezer Kanal, Ph.D. >>>> Postdoctoral Fellow >>>> Center for the Neural Basis of Cognition >>>> Carnegie Mellon University >>>> 4400 Fifth Ave, Suite 110A >>>> Pittsburgh PA 15213 >>>> P: 412-268-4115 >>>> F: 412-268-5060 >>>> >>>> >>>> _______________________________________________ >>>> fieldtrip mailing list >>>> fieldtrip at donders.ru.nl >>>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >>> >>> _______________________________________________ >>> fieldtrip mailing list >>> fieldtrip at donders.ru.nl >>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From semrich at wisc.edu Tue Aug 30 21:49:03 2011 From: semrich at wisc.edu (Stephen Emrich) Date: Tue, 30 Aug 2011 14:49:03 -0500 Subject: [FieldTrip] depsamplesregrT help Message-ID: Dear fieldtrip users and developers, I am attempting to use the depsamplesregrT statistic, but have encountered some problems. The documentation also appears to be incomplete. I am attempting to regress the outputs of ft_freqanalysis for 3 conditions against some values (the IVs). From what I can tell, depsamplesregrT should do this. Some issues I'm running into: cvar: cvar is not documented in this function, but i'm assuming from some searches that the cvar should be another row of values that actually correspond to the regressors (the IVs). If I attempt to run it without specifying cvar, I get an error. unitselved: even if I include the cvar in the design matrix, there is an error that 'unitselved' is undefined. I can't find any documentation on what this variable is or what the values should be. Any help would be appreciated. S- From jonas at obleser.de Wed Aug 31 12:10:40 2011 From: jonas at obleser.de (Jonas Obleser) Date: Wed, 31 Aug 2011 12:10:40 +0200 Subject: [FieldTrip] depsamplesregrT help Message-ID: <67EC1721-BF53-450B-A1DB-67F16F4E2CC7@obleser.de> Hi Stephen, from my understanding (greatly facilitated by Eric Maris a year ago or so), depsamplesregrT does only test for linear increases or decreases in your data, depending on the sequence in which you input your data conditions. For example, freqstatistics(cfg, cond1{:}, cond2{:}, cond3{:}) will result in positive clusters for data increasing (quasi-)linearly from cond1 to cond3, and negative clusters for data behaving vice versa. For regression/correlation with an independent (e.g., behavioural) regressor, we use a statfun Nathan Weisz (also active on the list here) once wrote and kindly shared with us. Hope this helps somewhat. Best wishes, J. Dr. Jonas Obleser Auditory Cognition Group Max Planck Institute for Human Cognitive and Brain Sciences Leipzig, Germany (p) +49 (0)341 9940 114 (e) obleser at cbs.mpg.de From jm.horschig at donders.ru.nl Wed Aug 31 13:02:52 2011 From: jm.horschig at donders.ru.nl (=?ISO-8859-1?Q?=22J=F6rn_M=2E_Horschig=22?=) Date: Wed, 31 Aug 2011 13:02:52 +0200 Subject: [FieldTrip] Issues with neighbours in ft_timelockstatistics In-Reply-To: References: Message-ID: <4E5E14DC.8050000@donders.ru.nl> Hi Anna, As you might have read, we recently updated how neighbours are computed and stored, and that you have to specify your neighbours manually now. Could you let me know how you defined the neighbours and how many channels your data has? Maybe there is a mismatch between the neighbour struct and your data in number of sensors, but that's just a wild guess. Anyhow, the result might look different depending on how you define your neighbours. Anyway, if you provide more information I can look into this and see how to fix it. Best, Jörn On 8/30/2011 5:46 PM, Anna Lambrechts wrote: > Dear fieldtrip community, > > Until about two weeks ago I was using a batch to process group > comparison statistics on MEG data using ft_timelockstatistics. A week > later, the same batch was not working anymore, even on the same data. > > The error I get is the following: > > *??? Improper index matrix reference. > > Error in ==> clusterstat>makechanneighbstructmat at 520 > [seld] = match_str(cfg.channel, cfg.neighbours(chan).label); > > Error in ==> clusterstat at 60 > channeighbstructmat = makechanneighbstructmat(cfg); > > Error in ==> statistics_montecarlo at 320 > [stat, cfg] = clusterstat(cfg, statrand, statobs,'issource',issource); > > Error in ==> statistics_wrapper at 290 > [stat, cfg] = statmethod(cfg, dat, design, 'issource',issource); > > Error in ==> ft_timelockstatistics at 124 > [stat, cfg] = statistics_wrapper(cfg, varargin{:}); > > Error in ==> myfunction_stats at 59 > gpstat = ft_timelockstatistics(cfg, data1, data2); > > *It seems the error concerns the 'neighbours.mat' file reading, even > though I am not sure about it. When I delete the two last lines in > this file (to match the size of matrix wanted) the error disappears, > but the processing is incomplete and false (or at least it does not > match the results I had before). > > Has anyone encounter the same issue? > Did you do any update recently that might explain this problem, and do > you have any idea how to remedy to it? > > Best wishes, > Anna. > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip -- Jörn M. Horschig PhD Student Donders Institute for Brain, Cognition and Behaviour Centre for Cognitive Neuroimaging Radboud University Nijmegen Neuronal Oscillations Group P.O. Box 9101 NL-6500 HB Nijmegen The Netherlands Contact: E-Mail: jm.horschig at donders.ru.nl Tel: +31-(0)24-36-68493 Web: http://www.ru.nl/donders Visiting address: Trigon, room 2.30 Kapittelweg 29 NL-6525 EN Nijmegen The Netherlands -------------- next part -------------- An HTML attachment was scrubbed... URL: From anna.lambrechts at gmail.com Wed Aug 31 13:40:33 2011 From: anna.lambrechts at gmail.com (Anna Lambrechts) Date: Wed, 31 Aug 2011 13:40:33 +0200 Subject: [FieldTrip] fieldtrip Digest, Vol 9, Issue 30 In-Reply-To: References: Message-ID: Hello, I have written yesterday about troubles while running ft_timelockstatistics. As it happens now, this seems to be a compatibility issue between matlab 2011 and the current fieldtrip version (which you can solve temporarily by using older versions of both). Best wishes, Anna. 2011/8/30 > Send fieldtrip mailing list submissions to > fieldtrip at donders.ru.nl > > To subscribe or unsubscribe via the World Wide Web, visit > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > or, via email, send a message with subject or body 'help' to > fieldtrip-request at donders.ru.nl > > You can reach the person managing the list at > fieldtrip-owner at donders.ru.nl > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of fieldtrip digest..." > > > Today's Topics: > > 1. Call for Papers :: NIPS 2011 Workshop on Interpretable > Decoding of Higher Cognitive States from Neural Data (Irina Simanova) > 2. Issues with neighbours in ft_timelockstatistics (Anna Lambrechts) > 3. Re: assessing significance in using ft_timelockanalysis > results (Kanal Eliezer) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Tue, 30 Aug 2011 14:22:31 +0200 > From: Irina Simanova > To: Email discussion list for the FieldTrip project > > Subject: [FieldTrip] Call for Papers :: NIPS 2011 Workshop on > Interpretable Decoding of Higher Cognitive States from Neural Data > Message-ID: <2E7CF2CC-C7AF-4937-935E-D0337E8445C8 at mpi.nl> > Content-Type: text/plain; charset="us-ascii"; Format="flowed"; > DelSp="yes" > > Interpretable Decoding of Higher Cognitive States from Neural Data > > NIPS 2011 Workshop, Dec 16 or 17, 2011, Granada, Spain > > (Please feel free to distribute the CFP to all the interested persons > and groups.) > > Overview > > Over recent years, machine learning methods have become a crucial > analytical tool in cognitive neuroscience (see reviews by Formisano et > al., 2008; Pereira et al., 2009). Decoding techniques have > dramatically increased the sensitivity of experiments, and so also the > subtlety of cognitive questions that can be asked. At the same time > the mental phenomena being studied are moving beyond lower-level > perceptual and motor processes which are directly grounded in external > measurable realities. > > Decoding higher cognition and interpreting the learned behaviour of > the classifiers used pose unique challenges, as these psychological > states are complex, fast-changing and often ill-defined. Contemporary > machine learning methods deal well with the small numbers of cases, > and high numbers of co-linear dimensions typical of neural data, and > are generally optimized to maximize classification performance, rather > than to enable meaningful interpretation of the features they learn > from. And indeed recent work has succeeded to decode psychological > phenomena including visual object recognition (e.g. Kriegeskorte et > al., 2008; Connolly et al., 2011), perceptual interpretation of sounds > (Staeren et al., 2009), lexical semantics (Mitchell et al., 2008; > Siminova et al., 2010; Devereux et al., 2010; Murphy et al., 2011), > decision making during game playing (Xiang et al., 2009) and the > process of mental arithmetic (Anderson et al., 2008). But for the > cognitive scientists who use these methods, the primary question is > often not "how much" but rather "how" and "why" the patterns of neural > activity identified by a machine learning algorithm encode particular > cognitive processes. > > The aim of this workshop is therefore to 1) discuss the achievements > and problems of the decoding of high-level cognitive states, and 2) > explore the use of machine learning methodologies and other > computational models that enable such cognitive interpretation of > neural recordings of different modalities. Advances in this field > require close collaboration between machine learning experts, > neuroscientists and cognitive scientists. Thus, this workshop is > highly interdisciplinary and will aim to attract submissions also from > outside the existing NIPS community. By stimulating discussions among > experts in the different fields, the workshop seeks to generate novel > insights and new directions for research. > > Topics of interest > > The field requires techniques that are capable of taking advantage of > spatially distributed patterns in the brain, that are separated in > space but coordinated in their activity. Methods should also be > sensitive to the fine-grained temporal patterns of multiple processes > - which may proceed in a serial fashion, overlapping or in parallel > with each-other, or in multiple passes with bidirectional information > flows. Different recording modalities have distinctive advantages: > fMRI provides very fine millimetre-level localisation in the brain but > poor temporal resolution, while EEG and MEG have millisecond temporal > resolution at the cost of spatial resolution. Ideally machine learning > methods would be able to meaningfully combine complementary > information from these different neuroimaging techniques (see e.g. De > Martino et al., 2010). Moreover, as the processes underlying higher > cognition are so complex, methods should be able to disentangle even > tightly linked and confounded subprocesses. Finally, general use > algorithms that could induce latent dimensions from neural data, and > so reveal the "hidden" psychological states, would be a dramatic > advance on current hypothesis-driven analytical paradigms. Originality > of approach is encouraged and submissions on any related > methodological approach are welcomed, such as: > > - Interpreting spatial and temporal location of selected features and > their weights > - Discovering "hidden" or "latent" cognitive representations > - Disentangling confounded processes and representations > - Comparing or combining data from recording modalities (e.g. fMRI, > EEG, structural MRI, DTI, MEG, NIRS, EcOG, single cell recordings) > - Fuzzy and partial classifications > - Unaligned or incommensurate feature spaces and data representation > > As noted above, the complexity of higher cognition poses challenges. > To take language comprehension as an example, speech is received at > 3-4 words; acoustic, semantic and syntactic processing can occur in > parallel; and the form of underlying representations (sentence > structures, conceptual descriptions) remains controversial. We welcome > submissions dealing with any high-level cognitive functions that > exhibit similar complexity, for instance: > > - Knowledge representation and concepts > - Language and communication > - Understanding visual and auditory experience > - Memory and learning > - Reasoning and problem solving > - Decision making and executive control > > Submissions > > Authors are invited to submit full papers on original, unpublished > work in the topic area of this workshop via the NIPS 2011 submission > site at https://cmt.research.microsoft.com/NIPS2011/Default.aspx. > Submissions should be formatted using the NIPS 2011 stylefiles, with > blind review and not exceeding 8 pages plus an extra page for > references. Author and submission information can be found at > http://nips.cc/PaperInformation/AuthorSubmissionInstructions > . The stylefiles are available at > http://nips.cc/PaperInformation/StyleFiles > . Each submission will be reviewed at least by two members of the > programme committee. Accepted papers will be published in the workshop > proceedings. Dual submissions to the main NIPS 2011 conference and > this workshop are allowed; if you submit to the main session, indicate > this when you submit to the workshop. If your paper is accepted for > the main session, you should withdraw your paper from the workshop > upon notification by the main session. > > Important Dates > > - Aug 30, 2011: Call for papers > - Sep 23, 2011: Deadline for submission of workshop papers > - Oct 15, 2011: Notification of acceptance > - Oct 31, 2011: Camera-ready papers due > - Dec 16 or 17, 2011: Workshop date > > Links > > - NIPS 2011 website: http://nips.cc/Conferences/2011/ > - Workshop website: https://sites.google.com/site/decodehighcogstate > - Call for Papers: https://sites.google.com/site/decodehighcogstate/cfp/ > > Kind regards, > The Workshop Organizers, > Kai-min Kevin Chang, Anna Korhonen, Irina Simanova, Brian Murphy > > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: < > http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20110830/27b062df/attachment-0001.html > > > > ------------------------------ > > Message: 2 > Date: Tue, 30 Aug 2011 17:46:52 +0200 > From: Anna Lambrechts > To: fieldtrip at donders.ru.nl > Subject: [FieldTrip] Issues with neighbours in ft_timelockstatistics > Message-ID: > > > Content-Type: text/plain; charset="iso-8859-1" > > Dear fieldtrip community, > > Until about two weeks ago I was using a batch to process group comparison > statistics on MEG data using ft_timelockstatistics. A week later, the same > batch was not working anymore, even on the same data. > > The error I get is the following: > > *??? Improper index matrix reference. > > Error in ==> clusterstat>makechanneighbstructmat at 520 > [seld] = match_str(cfg.channel, cfg.neighbours(chan).label); > > Error in ==> clusterstat at 60 > channeighbstructmat = makechanneighbstructmat(cfg); > > Error in ==> statistics_montecarlo at 320 > [stat, cfg] = clusterstat(cfg, statrand, statobs,'issource',issource); > > Error in ==> statistics_wrapper at 290 > [stat, cfg] = statmethod(cfg, dat, design, 'issource',issource); > > Error in ==> ft_timelockstatistics at 124 > [stat, cfg] = statistics_wrapper(cfg, varargin{:}); > > Error in ==> myfunction_stats at 59 > gpstat = ft_timelockstatistics(cfg, data1, data2); > > *It seems the error concerns the 'neighbours.mat' file reading, even though > I am not sure about it. When I delete the two last lines in this file (to > match the size of matrix wanted) the error disappears, but the processing > is > incomplete and false (or at least it does not match the results I had > before). > > Has anyone encounter the same issue? > Did you do any update recently that might explain this problem, and do you > have any idea how to remedy to it? > > Best wishes, > Anna. > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: < > http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20110830/f81c8aad/attachment-0001.html > > > > ------------------------------ > > Message: 3 > Date: Tue, 30 Aug 2011 11:58:34 -0400 > From: Kanal Eliezer > To: Email discussion list for the FieldTrip project > > Subject: Re: [FieldTrip] assessing significance in using > ft_timelockanalysis results > Message-ID: <8CC6BA25-9A03-4630-B940-486BE2B9EE92 at cmu.edu> > Content-Type: text/plain; charset=us-ascii > > Hello Eric - > > Thanks for the response. It looks like the fixed vs random analysis is > exactly what I'm referring to. From what I understood, it looks like the > difference really only shows up in the variance of the resultant > distribution; with a random, the variance also takes into account the > betwee-subjects variance. Is there a way to specify whether I want to do a > fixed or random effects analysis in FieldTrip when I'm running > ft_timelockgrandaverage or ft_freqgrandaverage? Thanks! > > Elli > > p.s. - In case anyone else is trying to figure this out, chapter 12 of > Friston's book "Statistical Parametric Mapping" does an excellent job > explaining the difference between fixed and random analyses, as well as how > to implement it algorithmically. > > > > On Aug 26, 2011, at 3:00 PM, Eric Maris wrote: > > > Hi Kanal, > > > >> The event related statistics tutorial > >> (http://fieldtrip.fcdonders.nl/tutorial/eventrelatedstatistics) talks > > about > >> assessing significance parametrically by running t-tests on pooled > >> timelockanalysis data. My question is, does the fact that the averages > > were > >> created from N trials make a difference? If I'm condition A has twelve > >> averages and condition B has another twelve, and each average contains > 70 > >> trials, is there a way to "inform" the statistical test that the power > in > > this > >> dataset is greater than 24? Is this only possible if I run the t-test > > comparing > >> each set of 840 (70*12) trials? > >> > >> I'm also curious whether this is possible with non-parametric analyses, > as > >> well. Thanks - > > > > In an analysis over subjects (called random-effects analysis in the fMRI > > literature), "informing" the statistical test about the number of trials > per > > condition only makes sense if this number is different for the two > > conditions. I propose that you have a look the fMRI papers that deal with > > the issue of fixed-versus-random effect analyses. The conceptual issues > > involved are the same in fMRI and electrophysiology. > > > > > > Best, > > > > Eric Maris > > > > > > > > > >> > >> Elli Kanal > >> > >> > >> -------------------- > >> Eliezer Kanal, Ph.D. > >> Postdoctoral Fellow > >> Center for the Neural Basis of Cognition > >> Carnegie Mellon University > >> 4400 Fifth Ave, Suite 110A > >> Pittsburgh PA 15213 > >> P: 412-268-4115 > >> F: 412-268-5060 > >> > >> > >> _______________________________________________ > >> fieldtrip mailing list > >> fieldtrip at donders.ru.nl > >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > _______________________________________________ > > fieldtrip mailing list > > fieldtrip at donders.ru.nl > > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > > ------------------------------ > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > End of fieldtrip Digest, Vol 9, Issue 30 > **************************************** > -------------- next part -------------- An HTML attachment was scrubbed... URL: From amelie.serpollet at cea.fr Wed Aug 3 10:30:38 2011 From: amelie.serpollet at cea.fr (=?iso-8859-1?Q?SERPOLLET_Am=E9lie_228173?=) Date: Wed, 3 Aug 2011 08:30:38 +0000 Subject: [FieldTrip] delay with realtime buffer Message-ID: <04B8DFAA8CFBED46A48C995E5BB1CD459A39@EXDAG0-B0.intra.cea.fr> Dear Fieldtrip users and developers, Using Fieldtrip buffer for a BCI loop (in realtime), I want to measure the delay introduced by the use of the buffer. I use two simple programs to measure it : the first one sends data to a buffer and writes CPU time in a file when a rising edge was detected in the signal sent ; the other program requests the data and also writes CPU time in a file when the same rising edge is detected in the signal received. I checked that the delay introduced by the detection and file writing is insignificant (0.5 to 5 µs). On my computer, with 32 channels signal and chunks of 8 to 32 points, I measured very variables delays, also after deactivating the anti-virus : 1ms to more than 45ms. Do you know what it is due to, if I can reduce it or if I can make it less variable ? Thank you in advance. Amélie -------------- next part -------------- An HTML attachment was scrubbed... URL: From Don.Rojas at ucdenver.edu Wed Aug 3 17:47:56 2011 From: Don.Rojas at ucdenver.edu (Rojas, Don) Date: Wed, 3 Aug 2011 09:47:56 -0600 Subject: [FieldTrip] ft_volumesegment question Message-ID: <621FD28E-EB95-4966-9612-B64308F79074@ucdenver.edu> Dear Fieldtrippers, I use FieldTrip and SPM8 together quite frequently. I noticed recently that the segmentation output (gray, white and csf tpm) of ft_volumesegment (c1, c2, c3) does not align properly to the input mri scan. This is not the case when doing the same segmentation in SPM8. I'm using the 20110725 version of Fieldtrip and have mad certain that it and not the one in spm8 is used. My cfg and command is as follows: cfg.output = {'brain' 'skull' 'scalp','tpm'}; cfg.spmversion = 'spm8'; cfg.name = 'test'; cfg.write = 'yes'; cfg.coordsys = 'spm'; seg = ft_volumesegment(cfg,mri); Looking at the code for ft_volumesegment, it appears that a permutation of the mri dimensions (swapping 2 and 3) is done on line 279: [mri,permutevec,flipflags] = align_ijk2xyz(mri); The mri.transform field is changed accordingly. However, the original.transform is put into the headers for the c1, c2 and c3 output (line 399). I think that this is not correct because the segmentation output has the permuted dimensions, not the dimensions of the original mri. So, the correct transform should be the mri.transform after align_ijk2xyz I think. Or maybe this is intentional behavior and I'm missing something. It can easily be checked, however, using Check Reg, in SPM8, that the output tpm do not align correctly with the original mri. If I manually write the mri.transform (from align_ijk2xyz) to the tpm output, however, they do align correctly. Best, Don ----------------------- Don Rojas, Ph.D. Associate Professor of Psychiatry U. of Colorado Denver Anschutz Medical Campus Director, UCD Magnetoencephalography Lab 13001 E. 17th Pl F546 Aurora, CO 80045 303-724-4994 -------------- next part -------------- An HTML attachment was scrubbed... URL: From ekanal at cmu.edu Wed Aug 3 23:29:52 2011 From: ekanal at cmu.edu (Kanal Eliezer) Date: Wed, 3 Aug 2011 17:29:52 -0400 Subject: [FieldTrip] timelock grand average can't average gradiometers? Message-ID: <2788E298-FE53-4F74-9C27-FDC29C0DDCBF@cmu.edu> Hello all - I'm trying to average a number of datasets collected from a Neuromag scanner. This dataset contains both magnetometer and gradiometer data. I'm trying to use ft_timelockgrandaverage to average the cleaned data, and I'm getting the following error message: > Warning: discarding gradiometer information because it cannot be averaged My question is, why not? My gradiometer data isn't different than the magnetometer data. Even more confusing, when looking in the code, it looks like it's averaging it anyway, and the resulting dataset seems to contain averaged grad data. What's the purpose of this error? Thanks, Eliezer Kanal From eelke.spaak at donders.ru.nl Thu Aug 4 08:03:32 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Thu, 4 Aug 2011 08:03:32 +0200 Subject: [FieldTrip] timelock grand average can't average gradiometers? In-Reply-To: <2788E298-FE53-4F74-9C27-FDC29C0DDCBF@cmu.edu> References: <2788E298-FE53-4F74-9C27-FDC29C0DDCBF@cmu.edu> Message-ID: Hi Eli, The warning message you mention regards the information about the gradiometer positions, i.e. the information in the data.grad field. The MEG-data recorded by the gradiometer sensors will still be averaged. I understand the confusion, though, and agree that the message could be a bit clearer. Best, Eelke 2011/8/3 Kanal Eliezer : > Hello all - > > I'm trying to average a number of datasets collected from a Neuromag scanner. This dataset contains both magnetometer and gradiometer data. I'm trying to use ft_timelockgrandaverage to average the cleaned data, and I'm getting the following error message: > >> Warning: discarding gradiometer information because it cannot be averaged > > My question is, why not? My gradiometer data isn't different than the magnetometer data. Even more confusing, when looking in the code, it looks like it's averaging it anyway, and the resulting dataset seems to contain averaged grad data. What's the purpose of this error? > > Thanks, > Eliezer Kanal > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From anna.lambrechts at gmail.com Thu Aug 4 11:17:46 2011 From: anna.lambrechts at gmail.com (Anna Lambrechts) Date: Thu, 4 Aug 2011 11:17:46 +0200 Subject: [FieldTrip] fieldtrip Digest, Vol 8, Issue 40 In-Reply-To: References: Message-ID: Hi, thank you for your first answer. Actually my question is really on the right way to implement a group comparison in fieldtrip (patient vs control group). Is it ok to run an independent t-test with the following design: design = zeros(2,n_group1+n_group2); design(2,:) = 1; design(1,1:n_group1) = 1:n_group1; design(1,n_group1+1:end) = 1:n_group2; A subsequent question is whether it is possible to implement a mixed design or we always have to proceed by combining conditions and compare it between groups. Cheers, Anna. > Message: 2 > Date: Sat, 30 Jul 2011 20:45:12 +0200 > From: "Eric Maris" > To: "'Email discussion list for the FieldTrip project'" > > Subject: Re: [FieldTrip] Group comparison statistics - ERF study > Message-ID: <03a601cc4ee8$d152b760$73f82620$@maris at donders.ru.nl> > Content-Type: text/plain; charset="us-ascii" > > Hi Anna, > > > > You can compare the two groups with respect to any linear combination that > can be formed using the observations in the 2 (conditions) x 2 (response > types) within-UO design. For testing interaction effects, these linear > combinations are the usual contrasts. > > > > > > Best, > > > > Eric Maris > > > > > > dr. Eric Maris > Donders Institute for Brain, Cognition and Behavior > > Radboud University > P.O. Box 9104 > 6500 HE Nijmegen > The Netherlands > T:+31 24 3612651 > Mobile: 06 39584581 > > F:+31 24 3616066 > mailto:e.maris at donders.ru.nl > > http://www.nphyscog.com/ > > > > > > > > From: fieldtrip-bounces at donders.ru.nl > [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Anna Lambrechts > Sent: donderdag 28 juli 2011 11:32 > To: fieldtrip at donders.ru.nl > Subject: [FieldTrip] Group comparison statistics - ERF study > > > > Hi, > > I am trying to run a group comparison analysis on event-related fields data > in a 2 (groups) x 2 (conditions) x 2 (response types) design. Is this > possible at all with any fieldtrip script? As far as I know implemented > statistics look at within-groups comparison. > > Thanks, > Anna. > > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: < > http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20110730/1f410391/attachment-0001.html > > > > ------------------------------ > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > End of fieldtrip Digest, Vol 8, Issue 40 > **************************************** > -------------- next part -------------- An HTML attachment was scrubbed... URL: From v.litvak at ion.ucl.ac.uk Thu Aug 4 12:39:53 2011 From: v.litvak at ion.ucl.ac.uk (Vladimir Litvak) Date: Thu, 4 Aug 2011 11:39:53 +0100 Subject: [FieldTrip] timelock grand average can't average gradiometers? In-Reply-To: References: <2788E298-FE53-4F74-9C27-FDC29C0DDCBF@cmu.edu> Message-ID: Dear Eelke, Actually there is a way to average sensors which is implemented in ft_average_sens. Perhaps it would be a good idea to use it in ft_timelockgrandaverage. Best, Vladimir On Thu, Aug 4, 2011 at 7:03 AM, Eelke Spaak wrote: > Hi Eli, > > The warning message you mention regards the information about the > gradiometer positions, i.e. the information in the data.grad field. > The MEG-data recorded by the gradiometer sensors will still be > averaged. > > I understand the confusion, though, and agree that the message could > be a bit clearer. > > Best, > Eelke > > 2011/8/3 Kanal Eliezer : >> Hello all - >> >> I'm trying to average a number of datasets collected from a Neuromag scanner. This dataset contains both magnetometer and gradiometer data. I'm trying to use ft_timelockgrandaverage to average the cleaned data, and I'm getting the following error message: >> >>> Warning: discarding gradiometer information because it cannot be averaged >> >> My question is, why not? My gradiometer data isn't different than the magnetometer data. Even more confusing, when looking in the code, it looks like it's averaging it anyway, and the resulting dataset seems to contain averaged grad data. What's the purpose of this error? >> >> Thanks, >> Eliezer Kanal >> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From eelke.spaak at donders.ru.nl Thu Aug 4 14:04:44 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Thu, 4 Aug 2011 14:04:44 +0200 Subject: [FieldTrip] timelock grand average can't average gradiometers? In-Reply-To: References: <2788E298-FE53-4F74-9C27-FDC29C0DDCBF@cmu.edu> Message-ID: Dear Vladimir, Thanks, I did not know about that function. However, I don't think ft_timelockgrandaverage should automatically average sensor positions; if the user wants this it is probably best if he/she does this explicitly. But, I will bring it up in the next FieldTrip dev team meeting, and see what others think (particularly Robert and Jan-Mathijs). Best, Eelke 2011/8/4 Vladimir Litvak : > Dear Eelke, > > Actually there is a way to average sensors which is implemented in > ft_average_sens. Perhaps it would be a good idea to use it in > ft_timelockgrandaverage. > > Best, > > Vladimir > > On Thu, Aug 4, 2011 at 7:03 AM, Eelke Spaak wrote: >> Hi Eli, >> >> The warning message you mention regards the information about the >> gradiometer positions, i.e. the information in the data.grad field. >> The MEG-data recorded by the gradiometer sensors will still be >> averaged. >> >> I understand the confusion, though, and agree that the message could >> be a bit clearer. >> >> Best, >> Eelke >> >> 2011/8/3 Kanal Eliezer : >>> Hello all - >>> >>> I'm trying to average a number of datasets collected from a Neuromag scanner. This dataset contains both magnetometer and gradiometer data. I'm trying to use ft_timelockgrandaverage to average the cleaned data, and I'm getting the following error message: >>> >>>> Warning: discarding gradiometer information because it cannot be averaged >>> >>> My question is, why not? My gradiometer data isn't different than the magnetometer data. Even more confusing, when looking in the code, it looks like it's averaging it anyway, and the resulting dataset seems to contain averaged grad data. What's the purpose of this error? >>> >>> Thanks, >>> Eliezer Kanal >>> >>> >>> _______________________________________________ >>> fieldtrip mailing list >>> fieldtrip at donders.ru.nl >>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >>> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From e.maris at donders.ru.nl Thu Aug 4 14:08:39 2011 From: e.maris at donders.ru.nl (Eric Maris) Date: Thu, 4 Aug 2011 14:08:39 +0200 Subject: [FieldTrip] fieldtrip Digest, Vol 8, Issue 40 In-Reply-To: References: Message-ID: <053f01cc529f$3efff020$bcffd060$@maris@donders.ru.nl> Hi Anna, thank you for your first answer. Actually my question is really on the right way to implement a group comparison in fieldtrip (patient vs control group). Is it ok to run an independent t-test with the following design: design = zeros(2,n_group1+n_group2); design(2,:) = 1; design(1,1:n_group1) = 1:n_group1; design(1,n_group1+1:end) = 1:n_group2; This is not correct. I advise you to have a look at the Fieldtrip tutorial on cluster-based permutation tests. A between-subjects experiment (e.g., patients versus controls) is formally equivalent to a between-trials experiment (explained in detail in the tutorial), and therefore should be analyzed in the same way. You can even copy-paste part of the tutorial code for the analysis of your between-subjects data. A subsequent question is whether it is possible to implement a mixed design or we always have to proceed by combining conditions and compare it between groups. I do not see the contrast between these two options. What is wrong with the second option? In fact, it is the option that is advocated in many statistics books (e.g., Maxwell & Delaney, Johnson & Wichren). Best, Eric Maris Cheers, Anna. Message: 2 Date: Sat, 30 Jul 2011 20:45:12 +0200 From: "Eric Maris" To: "'Email discussion list for the FieldTrip project'" Subject: Re: [FieldTrip] Group comparison statistics - ERF study Message-ID: <03a601cc4ee8$d152b760$73f82620$@maris at donders.ru.nl> Content-Type: text/plain; charset="us-ascii" Hi Anna, You can compare the two groups with respect to any linear combination that can be formed using the observations in the 2 (conditions) x 2 (response types) within-UO design. For testing interaction effects, these linear combinations are the usual contrasts. Best, Eric Maris dr. Eric Maris Donders Institute for Brain, Cognition and Behavior Radboud University P.O. Box 9104 6500 HE Nijmegen The Netherlands T:+31 24 3612651 Mobile: 06 39584581 F:+31 24 3616066 mailto:e.maris at donders.ru.nl http://www.nphyscog.com/ From: fieldtrip-bounces at donders.ru.nl [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Anna Lambrechts Sent: donderdag 28 juli 2011 11:32 To: fieldtrip at donders.ru.nl Subject: [FieldTrip] Group comparison statistics - ERF study Hi, I am trying to run a group comparison analysis on event-related fields data in a 2 (groups) x 2 (conditions) x 2 (response types) design. Is this possible at all with any fieldtrip script? As far as I know implemented statistics look at within-groups comparison. Thanks, Anna. -------------- next part -------------- An HTML attachment was scrubbed... URL: ------------------------------ _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip End of fieldtrip Digest, Vol 8, Issue 40 **************************************** -------------- next part -------------- An HTML attachment was scrubbed... URL: From drivolta81 at gmail.com Thu Aug 4 14:44:08 2011 From: drivolta81 at gmail.com (Davide Rivolta) Date: Thu, 4 Aug 2011 14:44:08 +0200 Subject: [FieldTrip] Potential bug in trl Message-ID: Dear all, I have noticed that the trial definition works well when begsample, ensample and offset have fixed values. I have however a specific trialfunction, where each trial has a different lenght. This works well when you specify a fixed offset, however, when even the offset has to change for each trial, Fieldtrip reports values until 0. It ignores time after 0. Might it be a bug? Thanks, Davide -------------- next part -------------- An HTML attachment was scrubbed... URL: From BelluscB at ninds.nih.gov Thu Aug 4 15:23:57 2011 From: BelluscB at ninds.nih.gov (Belluscio, Beth (NIH/NINDS) [E]) Date: Thu, 4 Aug 2011 09:23:57 -0400 Subject: [FieldTrip] time frequency analysis - basic question Message-ID: <794DFDD3C128EF44AC49ADC58FD0090C057456B617@NIHMLBX10.nih.gov> Hi all- When calling ft_freqanalysis you specify the foi limits and intervals - eg 5:5:30. The resulting data set appears to generate the power at discreet frequencies (5, 10, 15, 20, 25, 30). Does this actually reflect an estimate of the power over a range of frequencies (2.5-7.5, 7.5-12.5, 12.5-17.5,....) or the power at discreet frequencies with those in between not evaluated? Thanks for your help, Beth. Beth Belluscio MD-PhD Clinical Fellow Human Motor Control Section NINDS, NIH 301-402-3495 -------------- next part -------------- An HTML attachment was scrubbed... URL: From michael.wibral at web.de Thu Aug 4 15:55:46 2011 From: michael.wibral at web.de (Michael Wibral) Date: Thu, 4 Aug 2011 15:55:46 +0200 (CEST) Subject: [FieldTrip] Potential bug in trl Message-ID: <422581888.3319195.1312466146646.JavaMail.fmail@mwmweb086> An HTML attachment was scrubbed... URL: From drivolta81 at gmail.com Thu Aug 4 16:03:24 2011 From: drivolta81 at gmail.com (Davide Rivolta) Date: Thu, 4 Aug 2011 16:03:24 +0200 Subject: [FieldTrip] Potential bug in trl In-Reply-To: <422581888.3319195.1312466146646.JavaMail.fmail@mwmweb086> References: <422581888.3319195.1312466146646.JavaMail.fmail@mwmweb086> Message-ID: Thanks Michael, I wrote my own trialfun. The offset would be negative and would have the same lenght of difference between endsample-begsample. This, for each trial, would have a different value. It can be a bug. In fact, I tried to vhange the trialfun by leaving the same begsample and endsample, but I put a fixed offset (i.e., -100) and it worked. All the trial was in and not, like with the variable offset, until time = 0. Thanks, Davide On Thu, Aug 4, 2011 at 3:55 PM, Michael Wibral wrote: > Hi davide, > > I guess in such a case you would have to write your own trial function. > However, offset values larger than 0 mean, that the stimulus occured before > you start your piece of data - is that really what you wanted ? > > Michael > > > ------------------------------ > *Von:* "Davide Rivolta" > *Gesendet:* Aug 4, 2011 2:44:08 PM > *An:* fieldtrip at donders.ru.nl > *Betreff:* [FieldTrip] Potential bug in trl > > > Dear all, > > I have noticed that the trial definition works well when begsample, > ensample and offset have fixed values. > > I have however a specific trialfunction, where each trial has a different > lenght. > > This works well when you specify a fixed offset, however, when even the > offset has to change for each trial, Fieldtrip reports values until 0. It > ignores time after 0. > > > Might it be a bug? > > Thanks, > > Davide > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > -- Davide Rivolta, PhD -------------- next part -------------- An HTML attachment was scrubbed... URL: From jm.horschig at donders.ru.nl Fri Aug 5 09:13:23 2011 From: jm.horschig at donders.ru.nl (=?ISO-8859-1?Q?=22J=F6rn_M=2E_Horschig=22?=) Date: Fri, 05 Aug 2011 09:13:23 +0200 Subject: [FieldTrip] Potential bug in trl In-Reply-To: References: <422581888.3319195.1312466146646.JavaMail.fmail@mwmweb086> Message-ID: <4E3B9813.8070507@donders.ru.nl> Dear Davide, I never worked with trials of different length, however this should be straight forward when writing your own trialfun (just as Michael proposed). In most trialfuns, the trl matrix is constructed by concatenating the vectors begsample, endsample and offset. These vectors can have any values you want them to have. As a simple example, you can write a trialfun that just returns trl = [1 2 1; 400 500 100] (for your experiment, you of course want these values to be computed from your trigger values). Is your problem that such a thing does not work? It would be great if you give a concrete use-case, or e.g. attach your trialfun. Best, Jörn On 8/4/2011 4:03 PM, Davide Rivolta wrote: > Thanks Michael, > I wrote my own trialfun. > The offset would be negative and would have the same lenght of > difference between endsample-begsample. This, for each trial, would > have a different value. > It can be a bug. In fact, I tried to vhange the trialfun by leaving > the same begsample and endsample, but I put a fixed offset (i.e., > -100) and it worked. All the trial was in and not, like with the > variable offset, until time = 0. > Thanks, > Davide > > On Thu, Aug 4, 2011 at 3:55 PM, Michael Wibral > wrote: > > Hi davide, > > I guess in such a case you would have to write your own trial > function. However, offset values larger than 0 mean, that the > stimulus occured before you start your piece of data - is that > really what you wanted ? > > Michael > > > ------------------------------------------------------------------------ > *Von:* "Davide Rivolta" > > *Gesendet:* Aug 4, 2011 2:44:08 PM > *An:* fieldtrip at donders.ru.nl > *Betreff:* [FieldTrip] Potential bug in trl > > > Dear all, > I have noticed that the trial definition works well when > begsample, ensample and offset have fixed values. > I have however a specific trialfunction, where each trial has > a different lenght. > This works well when you specify a fixed offset, however, when > even the offset has to change for each trial, Fieldtrip > reports values until 0. It ignores time after 0. > Might it be a bug? > Thanks, > Davide > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > > -- > Davide Rivolta, PhD > > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip -- Jörn M. Horschig PhD Student Donders Institute for Brain, Cognition and Behaviour Centre for Cognitive Neuroimaging Radboud University Nijmegen Neuronal Oscillations Group P.O. Box 9101 NL-6500 HB Nijmegen The Netherlands Contact: E-Mail: jm.horschig at donders.ru.nl Tel: +31-(0)24-36-68493 Web: http://www.ru.nl/donders Visiting address: Trigon, room 2.30 Kapittelweg 29 NL-6525 EN Nijmegen The Netherlands -------------- next part -------------- An HTML attachment was scrubbed... URL: From jm.horschig at donders.ru.nl Fri Aug 5 09:29:38 2011 From: jm.horschig at donders.ru.nl (=?ISO-8859-1?Q?=22J=F6rn_M=2E_Horschig=22?=) Date: Fri, 05 Aug 2011 09:29:38 +0200 Subject: [FieldTrip] time frequency analysis - basic question In-Reply-To: <794DFDD3C128EF44AC49ADC58FD0090C057456B617@NIHMLBX10.nih.gov> References: <794DFDD3C128EF44AC49ADC58FD0090C057456B617@NIHMLBX10.nih.gov> Message-ID: <4E3B9BE2.4000300@donders.ru.nl> Dear Beth, From all my knowledge I gathered so far, I can try to answer your question (and hope that people from this list correct me if I am wrong): As often, the truth lies somewhere in between. The frequency smoothing is determined by the type and length of the taper you apply - this is similar to wavelet analysis. If you use multitapers, you can specify cfg.tapsmofrq, which determines the frequency band you want to smooth with. The FT freqanalysis method uses multiplication in the time domain instead of convolution in the frequency domain (which is equivalent). Any taper will always have a specific frequency spectrum itself, and thus it will not give you the value of a single frequency (though the output looks as if). So if you specify foi as [5 10 15], then the power at 5Hz will not be from 2.5 to 7.5Hz, but neither will it be at exactly and only 5Hz (btw, the exact center frequency might not be 5Hz, this depends on the (Rayleigh) frequency resolution of your signal). I do not know any more specifics concerning this, maybe the help of ft_freqanalysis can tell you some additonal things: http://fieldtrip.fcdonders.nl/reference/ft_freqanalysis Best, Jörn Btw, if this is not correct, else from the mailinglist should correct me asap ;) On 8/4/2011 3:23 PM, Belluscio, Beth (NIH/NINDS) [E] wrote: > > Hi all- > > When calling ft_freqanalysis you specify the foi limits and > intervals -- eg 5:5:30. The resulting data set appears to generate > the power at discreet frequencies (5, 10, 15, 20, 25, 30). Does this > actually reflect an estimate of the power over a range of frequencies > (2.5-7.5, 7.5-12.5, 12.5-17.5,....) or the power at discreet > frequencies with those in between not evaluated? > > Thanks for your help, > > Beth. > > Beth Belluscio MD-PhD > > Clinical Fellow > > Human Motor Control Section > > NINDS, NIH > > 301-402-3495 > > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip -- Jörn M. Horschig PhD Student Donders Institute for Brain, Cognition and Behaviour Centre for Cognitive Neuroimaging Radboud University Nijmegen Neuronal Oscillations Group P.O. Box 9101 NL-6500 HB Nijmegen The Netherlands Contact: E-Mail: jm.horschig at donders.ru.nl Tel: +31-(0)24-36-68493 Web: http://www.ru.nl/donders Visiting address: Trigon, room 2.30 Kapittelweg 29 NL-6525 EN Nijmegen The Netherlands -------------- next part -------------- An HTML attachment was scrubbed... URL: From drivolta81 at gmail.com Fri Aug 5 11:36:07 2011 From: drivolta81 at gmail.com (Davide Rivolta) Date: Fri, 5 Aug 2011 11:36:07 +0200 Subject: [FieldTrip] Potential bug in trl In-Reply-To: <4E3B9813.8070507@donders.ru.nl> References: <422581888.3319195.1312466146646.JavaMail.fmail@mwmweb086> <4E3B9813.8070507@donders.ru.nl> Message-ID: Dear Jörn, Thank you for your reply. I guess I have now figured it out. Thanks, Davide On Fri, Aug 5, 2011 at 9:13 AM, "Jörn M. Horschig" < jm.horschig at donders.ru.nl> wrote: > Dear Davide, > > I never worked with trials of different length, however this should be > straight forward when writing your own trialfun (just as Michael proposed). > In most trialfuns, the trl matrix is constructed by concatenating the > vectors begsample, endsample and offset. These vectors can have any values > you want them to have. As a simple example, you can write a trialfun that > just returns trl = [1 2 1; 400 500 100] (for your experiment, you of course > want these values to be computed from your trigger values). > Is your problem that such a thing does not work? It would be great if you > give a concrete use-case, or e.g. attach your trialfun. > > Best, > Jörn > > > On 8/4/2011 4:03 PM, Davide Rivolta wrote: > > Thanks Michael, > > I wrote my own trialfun. > > The offset would be negative and would have the same lenght of difference > between endsample-begsample. This, for each trial, would have a different > value. > > It can be a bug. In fact, I tried to vhange the trialfun by leaving the > same begsample and endsample, but I put a fixed offset (i.e., -100) and it > worked. All the trial was in and not, like with the variable offset, until > time = 0. > > Thanks, > > Davide > > On Thu, Aug 4, 2011 at 3:55 PM, Michael Wibral wrote: > >> Hi davide, >> >> I guess in such a case you would have to write your own trial function. >> However, offset values larger than 0 mean, that the stimulus occured before >> you start your piece of data - is that really what you wanted ? >> >> Michael >> >> >> ------------------------------ >> *Von:* "Davide Rivolta" >> *Gesendet:* Aug 4, 2011 2:44:08 PM >> *An:* fieldtrip at donders.ru.nl >> *Betreff:* [FieldTrip] Potential bug in trl >> >> >> Dear all, >> >> I have noticed that the trial definition works well when begsample, >> ensample and offset have fixed values. >> >> I have however a specific trialfunction, where each trial has a different >> lenght. >> >> This works well when you specify a fixed offset, however, when even the >> offset has to change for each trial, Fieldtrip reports values until 0. It >> ignores time after 0. >> >> >> Might it be a bug? >> >> Thanks, >> >> Davide >> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> > > > > -- > Davide Rivolta, PhD > > > > _______________________________________________ > fieldtrip mailing listfieldtrip at donders.ru.nlhttp://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > -- > Jörn M. Horschig > PhD Student > Donders Institute for Brain, Cognition and Behaviour > Centre for Cognitive Neuroimaging > Radboud University Nijmegen > Neuronal Oscillations Group > > P.O. Box 9101 > NL-6500 HB Nijmegen > The Netherlands > > Contact: > E-Mail: jm.horschig at donders.ru.nl > Tel: +31-(0)24-36-68493 > Web: http://www.ru.nl/donders > > Visiting address: > Trigon, room 2.30 > Kapittelweg 29 > NL-6525 EN Nijmegen > The Netherlands > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > -- Davide Rivolta, PhD -------------- next part -------------- An HTML attachment was scrubbed... URL: From BelluscB at ninds.nih.gov Fri Aug 5 16:20:11 2011 From: BelluscB at ninds.nih.gov (Belluscio, Beth (NIH/NINDS) [E]) Date: Fri, 5 Aug 2011 10:20:11 -0400 Subject: [FieldTrip] time frequency analysis - basic question In-Reply-To: <4E3B9BE2.4000300@donders.ru.nl> References: <794DFDD3C128EF44AC49ADC58FD0090C057456B617@NIHMLBX10.nih.gov> <4E3B9BE2.4000300@donders.ru.nl> Message-ID: <794DFDD3C128EF44AC49ADC58FD0090C057456B626@NIHMLBX10.nih.gov> Thanks, Jorn, that makes sense to me. From: "Jörn M. Horschig" [mailto:jm.horschig at donders.ru.nl] Sent: Friday, August 05, 2011 3:30 AM To: Email discussion list for the FieldTrip project Subject: Re: [FieldTrip] time frequency analysis - basic question Dear Beth, >From all my knowledge I gathered so far, I can try to answer your question (and hope that people from this list correct me if I am wrong): As often, the truth lies somewhere in between. The frequency smoothing is determined by the type and length of the taper you apply - this is similar to wavelet analysis. If you use multitapers, you can specify cfg.tapsmofrq, which determines the frequency band you want to smooth with. The FT freqanalysis method uses multiplication in the time domain instead of convolution in the frequency domain (which is equivalent). Any taper will always have a specific frequency spectrum itself, and thus it will not give you the value of a single frequency (though the output looks as if). So if you specify foi as [5 10 15], then the power at 5Hz will not be from 2.5 to 7.5Hz, but neither will it be at exactly and only 5Hz (btw, the exact center frequency might not be 5Hz, this depends on the (Rayleigh) frequency resolution of your signal). I do not know any more specifics concerning this, maybe the help of ft_freqanalysis can tell you some additonal things: http://fieldtrip.fcdonders.nl/reference/ft_freqanalysis Best, Jörn Btw, if this is not correct, else from the mailinglist should correct me asap ;) On 8/4/2011 3:23 PM, Belluscio, Beth (NIH/NINDS) [E] wrote: Hi all- When calling ft_freqanalysis you specify the foi limits and intervals - eg 5:5:30. The resulting data set appears to generate the power at discreet frequencies (5, 10, 15, 20, 25, 30). Does this actually reflect an estimate of the power over a range of frequencies (2.5-7.5, 7.5-12.5, 12.5-17.5,....) or the power at discreet frequencies with those in between not evaluated? Thanks for your help, Beth. Beth Belluscio MD-PhD Clinical Fellow Human Motor Control Section NINDS, NIH 301-402-3495 _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip -- Jörn M. Horschig PhD Student Donders Institute for Brain, Cognition and Behaviour Centre for Cognitive Neuroimaging Radboud University Nijmegen Neuronal Oscillations Group P.O. Box 9101 NL-6500 HB Nijmegen The Netherlands Contact: E-Mail: jm.horschig at donders.ru.nl Tel: +31-(0)24-36-68493 Web: http://www.ru.nl/donders Visiting address: Trigon, room 2.30 Kapittelweg 29 NL-6525 EN Nijmegen The Netherlands -------------- next part -------------- An HTML attachment was scrubbed... URL: From cas243 at georgetown.edu Sat Aug 6 17:21:47 2011 From: cas243 at georgetown.edu (Clara A. Scholl) Date: Sat, 6 Aug 2011 11:21:47 -0400 Subject: [FieldTrip] Extracting a timecourse from cluster of channels In-Reply-To: <4e34099a.c188e50a.3fc3.ffff8898SMTPIN_ADDED@mx.google.com> References: <4e34099a.c188e50a.3fc3.ffff8898SMTPIN_ADDED@mx.google.com> Message-ID: Thanks Eric! Do you think it's reasonable to take the intersection, or union, of channels over time in a given space-time cluster? Do you know of any publications that have dealt with this issue (choosing channels from a space-time cluster for timecourse extraction) that I could look at & reference? Thanks, Clara On Sat, Jul 30, 2011 at 9:31 AM, Eric Maris wrote: > Dear Clara, > >> I am using fieldtrip's cluster-based permutation test to identify a >> significant channel-time cluster showing an effect.  Now I would like >> to extract a timecourse averaged over these channels (for the purpose >> of comparing different conditions, not used to generate the cluster), >> but I'm not sure if it makes sense to do so because the channels >> belonging to the cluster change over time -- would the timecourse have >> contributions from different channels at different time points?  Do I >> need to limit my cluster to fixed channels across time? (And if so, >> how do I do that?) > > Yes, you should fix channels across time. The most important point in > selecting the channels is finding evidence for the fact that the channels > reflects a single source only. This is not a statistical issue, and > therefore there is not a p-value that can guide you in this choice. If you > have collected MEG data, the best evidence is a dipolar topography of your > effect. > > Best, > > Eric Maris > > >> >> Thanks! >> Clara >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From jan.schoffelen at donders.ru.nl Sat Aug 6 21:08:02 2011 From: jan.schoffelen at donders.ru.nl (jan-mathijs schoffelen) Date: Sat, 6 Aug 2011 21:08:02 +0200 Subject: [FieldTrip] ft_volumesegment question In-Reply-To: <621FD28E-EB95-4966-9612-B64308F79074@ucdenver.edu> References: <621FD28E-EB95-4966-9612-B64308F79074@ucdenver.edu> Message-ID: Dear Don, You're absolutely right. The correct tranfsormation matrix written to the file should be the mri.transform, not the original.transform. I'll fix this as soon as possible. Thanks for the debugging. Best wishes, Jan-Mathijs On Aug 3, 2011, at 5:47 PM, Rojas, Don wrote: > Dear Fieldtrippers, > > I use FieldTrip and SPM8 together quite frequently. I noticed recently that the segmentation output (gray, white and csf tpm) of ft_volumesegment (c1, c2, c3) does not align properly to the input mri scan. This is not the case when doing the same segmentation in SPM8. I'm using the 20110725 version of Fieldtrip and have mad certain that it and not the one in spm8 is used. My cfg and command is as follows: > > cfg.output = {'brain' 'skull' 'scalp','tpm'}; > cfg.spmversion = 'spm8'; > cfg.name = 'test'; > cfg.write = 'yes'; > cfg.coordsys = 'spm'; > seg = ft_volumesegment(cfg,mri); > > Looking at the code for ft_volumesegment, it appears that a permutation of the mri dimensions (swapping 2 and 3) is done on line 279: > > [mri,permutevec,flipflags] = align_ijk2xyz(mri); > > The mri.transform field is changed accordingly. However, the original.transform is put into the headers for the c1, c2 and c3 output (line 399). I think that this is not correct because the segmentation output has the permuted dimensions, not the dimensions of the original mri. So, the correct transform should be the mri.transform after align_ijk2xyz I think. Or maybe this is intentional behavior and I'm missing something. It can easily be checked, however, using Check Reg, in SPM8, that the output tpm do not align correctly with the original mri. If I manually write the mri.transform (from align_ijk2xyz) to the tpm output, however, they do align correctly. > > Best, > > Don > > ----------------------- > Don Rojas, Ph.D. > Associate Professor of Psychiatry > U. of Colorado Denver Anschutz Medical Campus > Director, UCD Magnetoencephalography Lab > 13001 E. 17th Pl F546 > Aurora, CO 80045 > 303-724-4994 > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip Dr. J.M. (Jan-Mathijs) Schoffelen Donders Institute for Brain, Cognition and Behaviour, Centre for Cognitive Neuroimaging, Radboud University Nijmegen, The Netherlands J.Schoffelen at donders.ru.nl Telephone: +31-24-3614793 -------------- next part -------------- An HTML attachment was scrubbed... URL: From michael.wibral at web.de Mon Aug 8 11:22:18 2011 From: michael.wibral at web.de (Michael Wibral) Date: Mon, 8 Aug 2011 11:22:18 +0200 (CEST) Subject: [FieldTrip] Potential bug in trl Message-ID: <434769618.3170746.1312795338080.JavaMail.fmail@mwmweb077> An HTML attachment was scrubbed... URL: From e.maris at donders.ru.nl Mon Aug 8 11:23:39 2011 From: e.maris at donders.ru.nl (Eric Maris) Date: Mon, 8 Aug 2011 11:23:39 +0200 Subject: [FieldTrip] Extracting a timecourse from cluster of channels In-Reply-To: References: <4e34099a.c188e50a.3fc3.ffff8898SMTPIN_ADDED@mx.google.com> Message-ID: <071401cc55ac$dea9af20$9bfd0d60$@maris@donders.ru.nl> Hi Clara, > Do you think it's reasonable to take the intersection, or union, of > channels over time in a given space-time cluster? > Do you know of any publications that have dealt with this issue > (choosing channels from a space-time cluster for timecourse > extraction) that I could look at & reference? I wish I could, but I cannot help you here. Your question belongs to the category "important but not answerable on a principled basis (at least for me)". Best, Eric Maris > Thanks, Clara > > On Sat, Jul 30, 2011 at 9:31 AM, Eric Maris wrote: > > Dear Clara, > > > >> I am using fieldtrip's cluster-based permutation test to identify a > >> significant channel-time cluster showing an effect.  Now I would like > >> to extract a timecourse averaged over these channels (for the purpose > >> of comparing different conditions, not used to generate the cluster), > >> but I'm not sure if it makes sense to do so because the channels > >> belonging to the cluster change over time -- would the timecourse have > >> contributions from different channels at different time points?  Do I > >> need to limit my cluster to fixed channels across time? (And if so, > >> how do I do that?) > > > > Yes, you should fix channels across time. The most important point in > > selecting the channels is finding evidence for the fact that the channels > > reflects a single source only. This is not a statistical issue, and > > therefore there is not a p-value that can guide you in this choice. If you > > have collected MEG data, the best evidence is a dipolar topography of your > > effect. > > > > Best, > > > > Eric Maris > > > > > >> > >> Thanks! > >> Clara > >> _______________________________________________ > >> fieldtrip mailing list > >> fieldtrip at donders.ru.nl > >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > _______________________________________________ > > fieldtrip mailing list > > fieldtrip at donders.ru.nl > > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From drivolta81 at gmail.com Mon Aug 8 12:17:45 2011 From: drivolta81 at gmail.com (Davide Rivolta) Date: Mon, 8 Aug 2011 12:17:45 +0200 Subject: [FieldTrip] Potential bug in trl In-Reply-To: <434769618.3170746.1312795338080.JavaMail.fmail@mwmweb077> References: <434769618.3170746.1312795338080.JavaMail.fmail@mwmweb077> Message-ID: Dear Michael, After your initial reply I realised that I made a (silly) mistake in the trialfun. It was not a bug indeed. My apologies for this, Davide On Mon, Aug 8, 2011 at 11:22 AM, Michael Wibral wrote: > Hi Davide, > > would you mind sharing your solution (that could potentially help others as > well)? > > Thanks, > Michael > > > > ------------------------------ > *Von:* "Davide Rivolta" > *Gesendet:* Aug 5, 2011 11:36:07 AM > *An:* "Email discussion list for the FieldTrip project" < > fieldtrip at donders.ru.nl> > *Betreff:* Re: [FieldTrip] Potential bug in trl > > > Dear Jörn, > > Thank you for your reply. I guess I have now figured it out. > > > Thanks, > Davide > > On Fri, Aug 5, 2011 at 9:13 AM, "Jörn M. Horschig" < > jm.horschig at donders.ru.nl> wrote: > >> Dear Davide, >> >> I never worked with trials of different length, however this should be >> straight forward when writing your own trialfun (just as Michael proposed). >> In most trialfuns, the trl matrix is constructed by concatenating the >> vectors begsample, endsample and offset. These vectors can have any values >> you want them to have. As a simple example, you can write a trialfun that >> just returns trl = [1 2 1; 400 500 100] (for your experiment, you of course >> want these values to be computed from your trigger values). >> Is your problem that such a thing does not work? It would be great if you >> give a concrete use-case, or e.g. attach your trialfun. >> >> Best, >> Jörn >> >> >> >> On 8/4/2011 4:03 PM, Davide Rivolta wrote: >> >> Thanks Michael, >> >> I wrote my own trialfun. >> >> The offset would be negative and would have the same lenght of difference >> between endsample-begsample. This, for each trial, would have a different >> value. >> >> It can be a bug. In fact, I tried to vhange the trialfun by leaving the >> same begsample and endsample, but I put a fixed offset (i.e., -100) and it >> worked. All the trial was in and not, like with the variable offset, until >> time = 0. >> >> Thanks, >> >> Davide >> >> On Thu, Aug 4, 2011 at 3:55 PM, Michael Wibral wrote: >> >>> Hi davide, >>> >>> I guess in such a case you would have to write your own trial function. >>> However, offset values larger than 0 mean, that the stimulus occured before >>> you start your piece of data - is that really what you wanted ? >>> >>> Michael >>> >>> >>> ------------------------------ >>> *Von:* "Davide Rivolta" >>> *Gesendet:* Aug 4, 2011 2:44:08 PM >>> *An:* fieldtrip at donders.ru.nl >>> *Betreff:* [FieldTrip] Potential bug in trl >>> >>> >>> Dear all, >>> >>> I have noticed that the trial definition works well when begsample, >>> ensample and offset have fixed values. >>> >>> I have however a specific trialfunction, where each trial has a different >>> lenght. >>> >>> This works well when you specify a fixed offset, however, when even the >>> offset has to change for each trial, Fieldtrip reports values until 0. It >>> ignores time after 0. >>> >>> >>> Might it be a bug? >>> >>> Thanks, >>> >>> Davide >>> >>> >>> _______________________________________________ >>> fieldtrip mailing list >>> fieldtrip at donders.ru.nl >>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> >> >> >> >> -- >> Davide Rivolta, PhD >> >> >> >> _______________________________________________ >> fieldtrip mailing listfieldtrip at donders.ru.nlhttp://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> >> >> >> >> -- >> Jörn M. Horschig >> PhD Student >> Donders Institute for Brain, Cognition and Behaviour >> Centre for Cognitive Neuroimaging >> Radboud University Nijmegen >> Neuronal Oscillations Group >> >> P.O. Box 9101 >> NL-6500 HB Nijmegen >> The Netherlands >> >> Contact: >> E-Mail: jm.horschig at donders.ru.nl >> Tel: +31-(0)24-36-68493 >> Web: http://www.ru.nl/donders >> >> Visiting address: >> Trigon, room 2.30 >> Kapittelweg 29 >> NL-6525 EN Nijmegen >> The Netherlands >> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > > -- > Davide Rivolta, PhD > > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > -- Davide Rivolta, PhD -------------- next part -------------- An HTML attachment was scrubbed... URL: From marco.dahmane at gmail.com Mon Aug 8 15:13:24 2011 From: marco.dahmane at gmail.com (Marco Dahmane) Date: Mon, 8 Aug 2011 14:13:24 +0100 Subject: [FieldTrip] databrowser and merged data Message-ID: Hi all, I have a rather straightforward question. I would like to be able to visualize the output of ft_timelockanalysis in butterfly mode using the databrowser function. However, since my data was made by merging several raw datasets together, the sampleinfo field was erased during the process, and databrowser doesn't like that (even though the timelock.time and timelock.avg fields are fine) What I would like to know is, since I have quite a lot of small datasets to be merged together, is there a way to make FT automatically reconstruct the sampleinfo field after merging two datasets together ? In other words, do I *have* to manually add the sampleinfo field at the end (and how come it is lost upon merging two datasets?)? And if yes, what is the easiest way to go about it? Many thanks, --Marco -------------- next part -------------- An HTML attachment was scrubbed... URL: From eelke.spaak at donders.ru.nl Mon Aug 8 15:36:11 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Mon, 8 Aug 2011 15:36:11 +0200 Subject: [FieldTrip] databrowser and merged data In-Reply-To: References: Message-ID: Dear Marco, Let me make three points in response to your question. (1) Actually, in general, timelock data structures should never contain sampleinfo, since sampleinfo describes which samples in the original recording your trials correspond to (I'm not sure about timelocked data with cfg.keeptrials='yes'). (2) However, I don't think that point (1) is relevant to your problem. More relevant is this: why do you want to use the databrowser to visualise timelocked data? The databrowser is meant to conveniently page through data with a big time axis. Usually, your timelocked data will not have a big time axis, but a small one. For this, we have the ft_singleplotER-function, or simply matlab's plot(). (3) When you provide data that lacks a sampleinfo field to a fieldtrip function that requires a sampleinfo field, the sampleinfo should be automatically constructed. (This is taken care of by ft_checkdata, which is also responsible for converting your timelocked data into raw data in case you pass it to the databrowser.) So, even if you decide that you really want to use the databrowser to display timelocked data, it should actually just work :) Are you using the latest version of fieldtrip? If so, could you provide me with some more details on the error and its circumstances? Best, Eelke 2011/8/8 Marco Dahmane : > Hi all, > I have a rather straightforward question. I would like to be able to > visualize the output of ft_timelockanalysis in butterfly mode using the > databrowser function. > However, since my data was made by merging several raw datasets together, > the sampleinfo field was erased during the process, and databrowser doesn't > like that (even though the timelock.time and timelock.avg fields are fine) > What I would like to know is, since I have quite a lot of small datasets to > be merged together, is there a way to make FT automatically reconstruct the > sampleinfo field after merging two datasets together ? In other words, do I > *have* to manually add the sampleinfo field at the end (and how come it is > lost upon merging two datasets?)? > And if yes, what is the easiest way to go about it? > Many thanks, > --Marco > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From marco.dahmane at gmail.com Mon Aug 8 16:08:11 2011 From: marco.dahmane at gmail.com (Marco Dahmane) Date: Mon, 8 Aug 2011 15:08:11 +0100 Subject: [FieldTrip] databrowser and merged data Message-ID: Hi Eelke, So in order :) 1) Fair enough. And I'm not using cfg.keeptrials='yes' anyway 2) I'm actually using databrowser because I find it really convenient to just remove and add channels, or see to which channel each line corresponds. I know I could do this using Matlab's plot, but I was just wondering why it wouldn't work with databrowser. 3) Ok this is weird then. Something that will probably help you, whenever I use databrowser on my timelocked data, I get this : Warning: the trial definition in the configuration is inconsistent with the actual data > In utilities/private/warning_once at 75 In utilities/private/fixsampleinfo at 51 In ft_checkdata at 579 In ft_databrowser at 155 Warning: failed to create sampleinfo field > In utilities/private/warning_once at 75 In utilities/private/fixsampleinfo at 83 In ft_checkdata at 579 In ft_databrowser at 155 So there must be something wrong right? And yes I'm using the latest FT... Many thanks, --Marco -------------- next part -------------- An HTML attachment was scrubbed... URL: From BelluscB at ninds.nih.gov Mon Aug 8 18:08:16 2011 From: BelluscB at ninds.nih.gov (Belluscio, Beth (NIH/NINDS) [E]) Date: Mon, 8 Aug 2011 12:08:16 -0400 Subject: [FieldTrip] power line noise reduction Message-ID: <794DFDD3C128EF44AC49ADC58FD0090C057456B635@NIHMLBX10.nih.gov> Hi all- I am evaluating power using ft_freqanalysis for resting state data. In the preprocessing step, I indicate to remove power line noise at 60 Hz as follows: cfg = []; cfg.dataset = 'resting.ds'; cfg.channel = {'MEG'}; cfg.continuous = 'yes'; cfg.detrend = 'yes'; cfg.dftfilter = 'yes'; cfg.dftfreq = [60 120]; dataCB24rest = ft_preprocessing(cfg) Then I evaluate the power at frequencies from 5 to 100 Hz as follows: cfg = []; cfg.method = 'mtmfft'; cfg.taper = 'hanning'; cfg.foi = 5:5:100; cfg.tapsmofrq = 4; cfg.output = 'pow'; freqCB24rest = ft_freqanalysis(cfg, dataCB24rest) When I visualize the result, I get a VERY large value for the power at 60 Hz in every channel. For example, see attached. What am I doing wrong? Thanks for your help, Beth. Beth Belluscio MD-PhD Clinical Fellow Human Motor Control Section NINDS, NIH 301-402-3495 -------------- next part -------------- A non-text attachment was scrubbed... Name: example.tif Type: image/tiff Size: 18858 bytes Desc: example.tif URL: From jan.schoffelen at donders.ru.nl Mon Aug 8 19:43:40 2011 From: jan.schoffelen at donders.ru.nl (jan-mathijs schoffelen) Date: Mon, 8 Aug 2011 19:43:40 +0200 Subject: [FieldTrip] power line noise reduction In-Reply-To: <794DFDD3C128EF44AC49ADC58FD0090C057456B635@NIHMLBX10.nih.gov> References: <794DFDD3C128EF44AC49ADC58FD0090C057456B635@NIHMLBX10.nih.gov> Message-ID: <0E61E958-F55D-4D1C-B3CB-7F25C9E3A63E@donders.ru.nl> Hi Beth, It seems from your approach, that you use a long stretch of data on which you apply the dftfilter. The dftfilter only works well, if the power line fluctuations are stationary in amplitude. Another way of stating it, is that the line noise really has to be a very narrowband line in the spectrum (relative to the epoch length). Typically, the dftfilter works well only up to an epoch length of ~ 10 seconds (sometimes even less if there are sources of non-stationary power line interference). In your case I would either chop up the data in shorter segments, or use a notch filter (or a lowpassfilter; I would rather go for the low frequency stuff to begin with in the resting state). Best, Jan-Mathijs On Aug 8, 2011, at 6:08 PM, Belluscio, Beth (NIH/NINDS) [E] wrote: > Hi all- > I am evaluating power using ft_freqanalysis for resting state data. > In the preprocessing step, I indicate to remove power line noise at 60 Hz as follows: > > cfg = []; > cfg.dataset = 'resting.ds'; > cfg.channel = {'MEG'}; > cfg.continuous = 'yes'; > cfg.detrend = 'yes'; > cfg.dftfilter = 'yes'; > cfg.dftfreq = [60 120]; > dataCB24rest = ft_preprocessing(cfg) > > Then I evaluate the power at frequencies from 5 to 100 Hz as follows: > > cfg = []; > cfg.method = 'mtmfft'; > cfg.taper = 'hanning'; > cfg.foi = 5:5:100; > cfg.tapsmofrq = 4; > cfg.output = 'pow'; > freqCB24rest = ft_freqanalysis(cfg, dataCB24rest) > > When I visualize the result, I get a VERY large value for the power at 60 Hz in every channel. For example, see attached. > > What am I doing wrong? > Thanks for your help, > Beth. > > > Beth Belluscio MD-PhD > Clinical Fellow > Human Motor Control Section > NINDS, NIH > 301-402-3495 > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip Dr. J.M. (Jan-Mathijs) Schoffelen Donders Institute for Brain, Cognition and Behaviour, Centre for Cognitive Neuroimaging, Radboud University Nijmegen, The Netherlands J.Schoffelen at donders.ru.nl Telephone: +31-24-3614793 From ekanal at cmu.edu Mon Aug 8 20:07:26 2011 From: ekanal at cmu.edu (Kanal Eliezer) Date: Mon, 8 Aug 2011 14:07:26 -0400 Subject: [FieldTrip] question about re-doing analysis Message-ID: <150C06A5-3B96-4F2C-B529-DB1079A36701@cmu.edu> Hello folks - Simple question here... I've been working with a dataset for a while, and it turns out that the [1 50] bandpass filter I applied earlier is removing some data that we're going to want to examine. My question is, what's the best way to go about this? For most of the data, the preprocessing consisted of discarding trials (a lot of noisy trials, unfortunately) and ICA. I've saved the data from the preprocessing stages, and my current thought is to write a script to load up each dataset, extract the "sampleinfo" field, pull that from the original raw file, apply any new filters, and then work with that. This would mitigate my needing to go through all the data and do the trial reject again (which would take weeks of work). Is there a better way to do this? Thanks - Eliezer Kanal From jan.schoffelen at donders.ru.nl Mon Aug 8 20:33:48 2011 From: jan.schoffelen at donders.ru.nl (jan-mathijs schoffelen) Date: Mon, 8 Aug 2011 20:33:48 +0200 Subject: [FieldTrip] question about re-doing analysis In-Reply-To: <150C06A5-3B96-4F2C-B529-DB1079A36701@cmu.edu> References: <150C06A5-3B96-4F2C-B529-DB1079A36701@cmu.edu> Message-ID: <80BC061E-C9FD-46B5-B3D7-289260D35533@donders.ru.nl> Dear Elli, > to load up each dataset, extract the "sampleinfo" field, pull that from the original raw file, apply any new filters, and then work with that. This would mitigate my needing to go through all the data and do the trial reject again (which would take weeks of work). Is there a better way to do this? Isn't this good enough, then ;-) ? JM Dr. J.M. (Jan-Mathijs) Schoffelen Donders Institute for Brain, Cognition and Behaviour, Centre for Cognitive Neuroimaging, Radboud University Nijmegen, The Netherlands J.Schoffelen at donders.ru.nl Telephone: +31-24-3614793 From ekanal at cmu.edu Mon Aug 8 20:58:08 2011 From: ekanal at cmu.edu (Kanal Eliezer) Date: Mon, 8 Aug 2011 14:58:08 -0400 Subject: [FieldTrip] question about re-doing analysis In-Reply-To: <80BC061E-C9FD-46B5-B3D7-289260D35533@donders.ru.nl> References: <150C06A5-3B96-4F2C-B529-DB1079A36701@cmu.edu> <80BC061E-C9FD-46B5-B3D7-289260D35533@donders.ru.nl> Message-ID: <024FF5CF-ECE6-44ED-9C35-2D7636D797CF@cmu.edu> Yes, that works... the only problem with this approach is that I'll have to redo ICA filtering. I guess I was just curious if there's a different way that everyone uses that I'm not familiar with. This is the first dataset I've analyzed using fieldtrip, so I'm just trying to make sure I'm "doing it right". Thanks - Elli On Aug 8, 2011, at 2:33 PM, jan-mathijs schoffelen wrote: > Dear Elli, > >> to load up each dataset, extract the "sampleinfo" field, pull that from the original raw file, apply any new filters, and then work with that. This would mitigate my needing to go through all the data and do the trial reject again (which would take weeks of work). Is there a better way to do this? > > Isn't this good enough, then ;-) ? > > JM > > > > Dr. J.M. (Jan-Mathijs) Schoffelen > Donders Institute for Brain, Cognition and Behaviour, > Centre for Cognitive Neuroimaging, > Radboud University Nijmegen, The Netherlands > J.Schoffelen at donders.ru.nl > Telephone: +31-24-3614793 > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From jan.schoffelen at donders.ru.nl Mon Aug 8 21:05:12 2011 From: jan.schoffelen at donders.ru.nl (jan-mathijs schoffelen) Date: Mon, 8 Aug 2011 21:05:12 +0200 Subject: [FieldTrip] question about re-doing analysis In-Reply-To: <024FF5CF-ECE6-44ED-9C35-2D7636D797CF@cmu.edu> References: <150C06A5-3B96-4F2C-B529-DB1079A36701@cmu.edu> <80BC061E-C9FD-46B5-B3D7-289260D35533@donders.ru.nl> <024FF5CF-ECE6-44ED-9C35-2D7636D797CF@cmu.edu> Message-ID: <6044F712-0B26-4637-B05B-9E89E9A7D206@donders.ru.nl> Hi Elli, Yes, that's a good point. What I sometimes do, is to also store the mixing matrix which comes out after ft_componentanalysis (i.e. the comp.topo). When you have this matrix, you can very cheaply re-ICA-filter the data. Yet, note that with different filter settings in the preprocessing, the ICA-components (and topographies) may also change. BW, JM On Aug 8, 2011, at 8:58 PM, Kanal Eliezer wrote: > Yes, that works... the only problem with this approach is that I'll have to redo ICA filtering. > > I guess I was just curious if there's a different way that everyone uses that I'm not familiar with. This is the first dataset I've analyzed using fieldtrip, so I'm just trying to make sure I'm "doing it right". Thanks - > > Elli > > > On Aug 8, 2011, at 2:33 PM, jan-mathijs schoffelen wrote: > >> Dear Elli, >> >>> to load up each dataset, extract the "sampleinfo" field, pull that from the original raw file, apply any new filters, and then work with that. This would mitigate my needing to go through all the data and do the trial reject again (which would take weeks of work). Is there a better way to do this? >> >> Isn't this good enough, then ;-) ? >> >> JM >> >> >> >> Dr. J.M. (Jan-Mathijs) Schoffelen >> Donders Institute for Brain, Cognition and Behaviour, >> Centre for Cognitive Neuroimaging, >> Radboud University Nijmegen, The Netherlands >> J.Schoffelen at donders.ru.nl >> Telephone: +31-24-3614793 >> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip Dr. J.M. (Jan-Mathijs) Schoffelen Donders Institute for Brain, Cognition and Behaviour, Centre for Cognitive Neuroimaging, Radboud University Nijmegen, The Netherlands J.Schoffelen at donders.ru.nl Telephone: +31-24-3614793 From ekanal at cmu.edu Mon Aug 8 21:19:42 2011 From: ekanal at cmu.edu (Kanal Eliezer) Date: Mon, 8 Aug 2011 15:19:42 -0400 Subject: [FieldTrip] question about re-doing analysis In-Reply-To: <6044F712-0B26-4637-B05B-9E89E9A7D206@donders.ru.nl> References: <150C06A5-3B96-4F2C-B529-DB1079A36701@cmu.edu> <80BC061E-C9FD-46B5-B3D7-289260D35533@donders.ru.nl> <024FF5CF-ECE6-44ED-9C35-2D7636D797CF@cmu.edu> <6044F712-0B26-4637-B05B-9E89E9A7D206@donders.ru.nl> Message-ID: <5B688DA3-CECE-42F5-9E0C-70F3FCB086FB@cmu.edu> That's a good point, I guess I'll have needed to re-do them anyway. Elli On Aug 8, 2011, at 3:05 PM, jan-mathijs schoffelen wrote: > Hi Elli, > > Yes, that's a good point. What I sometimes do, is to also store the mixing matrix which comes out after ft_componentanalysis (i.e. the comp.topo). When you have this matrix, you can very cheaply re-ICA-filter the data. Yet, note that with different filter settings in the preprocessing, the ICA-components (and topographies) may also change. > > > BW, JM > > > > On Aug 8, 2011, at 8:58 PM, Kanal Eliezer wrote: > >> Yes, that works... the only problem with this approach is that I'll have to redo ICA filtering. >> >> I guess I was just curious if there's a different way that everyone uses that I'm not familiar with. This is the first dataset I've analyzed using fieldtrip, so I'm just trying to make sure I'm "doing it right". Thanks - >> >> Elli >> >> >> On Aug 8, 2011, at 2:33 PM, jan-mathijs schoffelen wrote: >> >>> Dear Elli, >>> >>>> to load up each dataset, extract the "sampleinfo" field, pull that from the original raw file, apply any new filters, and then work with that. This would mitigate my needing to go through all the data and do the trial reject again (which would take weeks of work). Is there a better way to do this? >>> >>> Isn't this good enough, then ;-) ? >>> >>> JM >>> >>> >>> >>> Dr. J.M. (Jan-Mathijs) Schoffelen >>> Donders Institute for Brain, Cognition and Behaviour, >>> Centre for Cognitive Neuroimaging, >>> Radboud University Nijmegen, The Netherlands >>> J.Schoffelen at donders.ru.nl >>> Telephone: +31-24-3614793 >>> >>> >>> _______________________________________________ >>> fieldtrip mailing list >>> fieldtrip at donders.ru.nl >>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > Dr. J.M. (Jan-Mathijs) Schoffelen > Donders Institute for Brain, Cognition and Behaviour, > Centre for Cognitive Neuroimaging, > Radboud University Nijmegen, The Netherlands > J.Schoffelen at donders.ru.nl > Telephone: +31-24-3614793 > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From eelke.spaak at donders.ru.nl Tue Aug 9 11:00:27 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Tue, 9 Aug 2011 11:00:27 +0200 Subject: [FieldTrip] databrowser and merged data In-Reply-To: References: Message-ID: Hi Marco, The subfunction fixsampleinfo (used by ft_checkdata) attempts a number of things in order to create a sampleinfo field. My guess is that it detected a trial definition somewhere in your cfg "tree" (the data.cfg and data.cfg.previous fields) that was inconsistent with your data, and therefore refused to continue rebuilding a sampleinfo from scratch. I have now changed this function so that it will always reconstruct your sampleinfo, even if it detects an inconsistent trial definition in a previous configuration structure. So, it should now work :) At least, with tomorrow's update on the FTP server. Best, Eelke 2011/8/8 Marco Dahmane : > Hi Eelke, > So in order :) > 1) Fair enough. And I'm not using cfg.keeptrials='yes' anyway > 2) I'm actually using databrowser because I find it really convenient to > just remove and add channels, or see to which channel each line corresponds. > I know I could do this using Matlab's plot, but I was just wondering why it > wouldn't work with databrowser. > 3) Ok this is weird then. Something that will probably help you, whenever I > use databrowser on my timelocked data, I get this : > Warning: the trial definition in the configuration is inconsistent with the > actual data >> In utilities/private/warning_once at 75 >   In utilities/private/fixsampleinfo at 51 >   In ft_checkdata at 579 >   In ft_databrowser at 155 > Warning: failed to create sampleinfo field >> In utilities/private/warning_once at 75 >   In utilities/private/fixsampleinfo at 83 >   In ft_checkdata at 579 >   In ft_databrowser at 155 >  So there must be something wrong right? > And yes I'm using the latest FT... > Many thanks, > --Marco > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From jm.horschig at donders.ru.nl Tue Aug 9 12:21:45 2011 From: jm.horschig at donders.ru.nl (=?ISO-8859-1?Q?=22J=F6rn_M=2E_Horschig=22?=) Date: Tue, 09 Aug 2011 12:21:45 +0200 Subject: [FieldTrip] databrowser and merged data In-Reply-To: References: Message-ID: <4E410A39.2060809@donders.ru.nl> Hi Marco and Eelke, sampleinfo cannot be reconstructed after resampling the data, because there is no way to precisely relate your resampled samples to the original samples. My guess would be that you, Marco, did that and therefore encounter the problem you mentioned. If that was so, I am not sure whether fixsampleinfo should actually recompute the sampleinfo field, because it makes absolutely no sense to have a sampleinfo field in your resampled data that relates back samples in the original data. Anyway, if you did not resample your data, then fixsampleinfo should work. Could you let us know how you preprocessed your data? Best, Jörn On 8/9/2011 11:00 AM, Eelke Spaak wrote: > Hi Marco, > > The subfunction fixsampleinfo (used by ft_checkdata) attempts a number > of things in order to create a sampleinfo field. My guess is that it > detected a trial definition somewhere in your cfg "tree" (the data.cfg > and data.cfg.previous fields) that was inconsistent with your data, > and therefore refused to continue rebuilding a sampleinfo from > scratch. I have now changed this function so that it will always > reconstruct your sampleinfo, even if it detects an inconsistent trial > definition in a previous configuration structure. > > So, it should now work :) At least, with tomorrow's update on the FTP server. > > Best, > Eelke > > 2011/8/8 Marco Dahmane: >> Hi Eelke, >> So in order :) >> 1) Fair enough. And I'm not using cfg.keeptrials='yes' anyway >> 2) I'm actually using databrowser because I find it really convenient to >> just remove and add channels, or see to which channel each line corresponds. >> I know I could do this using Matlab's plot, but I was just wondering why it >> wouldn't work with databrowser. >> 3) Ok this is weird then. Something that will probably help you, whenever I >> use databrowser on my timelocked data, I get this : >> Warning: the trial definition in the configuration is inconsistent with the >> actual data >>> In utilities/private/warning_once at 75 >> In utilities/private/fixsampleinfo at 51 >> In ft_checkdata at 579 >> In ft_databrowser at 155 >> Warning: failed to create sampleinfo field >>> In utilities/private/warning_once at 75 >> In utilities/private/fixsampleinfo at 83 >> In ft_checkdata at 579 >> In ft_databrowser at 155 >> So there must be something wrong right? >> And yes I'm using the latest FT... >> Many thanks, >> --Marco >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip -- Jörn M. Horschig PhD Student Donders Institute for Brain, Cognition and Behaviour Centre for Cognitive Neuroimaging Radboud University Nijmegen Neuronal Oscillations Group P.O. Box 9101 NL-6500 HB Nijmegen The Netherlands Contact: E-Mail: jm.horschig at donders.ru.nl Tel: +31-(0)24-36-68493 Web: http://www.ru.nl/donders Visiting address: Trigon, room 2.30 Kapittelweg 29 NL-6525 EN Nijmegen The Netherlands From eelke.spaak at donders.ru.nl Tue Aug 9 12:28:21 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Tue, 9 Aug 2011 12:28:21 +0200 Subject: [FieldTrip] databrowser and merged data In-Reply-To: <4E410A39.2060809@donders.ru.nl> References: <4E410A39.2060809@donders.ru.nl> Message-ID: Hi Jörn, As you can read in my email, I actually changed fixsampleinfo so that it *does* reconstruct the sample info now :) I agree that sampleinfo does not make sense in the case of data that is (1) timelock-averaged and (2) combined from multiple datasets, but the issue here was to get the databrowser working on such data. Since the data used internally by the databrowser (after the call to ft_checkdata) is never exposed to the user, this seems like a good solution to me. Best, Eelke 2011/8/9 "Jörn M. Horschig" : > Hi Marco and Eelke, > > sampleinfo cannot be reconstructed after resampling the data, because there > is no way to precisely relate your resampled samples to the original > samples. My guess would be that you, Marco, did that and therefore encounter > the problem you mentioned. If that was so, I am not sure whether > fixsampleinfo should actually recompute the sampleinfo field, because it > makes absolutely no sense to have a sampleinfo field in your resampled data > that relates back samples in the original data. > Anyway, if you did not resample your data, then fixsampleinfo should work. > Could you let us know how you preprocessed your data? > > Best, > Jörn > > On 8/9/2011 11:00 AM, Eelke Spaak wrote: >> >> Hi Marco, >> >> The subfunction fixsampleinfo (used by ft_checkdata) attempts a number >> of things in order to create a sampleinfo field. My guess is that it >> detected a trial definition somewhere in your cfg "tree" (the data.cfg >> and data.cfg.previous fields) that was inconsistent with your data, >> and therefore refused to continue rebuilding a sampleinfo from >> scratch. I have now changed this function so that it will always >> reconstruct your sampleinfo, even if it detects an inconsistent trial >> definition in a previous configuration structure. >> >> So, it should now work :) At least, with tomorrow's update on the FTP >> server. >> >> Best, >> Eelke >> >> 2011/8/8 Marco Dahmane: >>> >>> Hi Eelke, >>> So in order :) >>> 1) Fair enough. And I'm not using cfg.keeptrials='yes' anyway >>> 2) I'm actually using databrowser because I find it really convenient to >>> just remove and add channels, or see to which channel each line >>> corresponds. >>> I know I could do this using Matlab's plot, but I was just wondering why >>> it >>> wouldn't work with databrowser. >>> 3) Ok this is weird then. Something that will probably help you, whenever >>> I >>> use databrowser on my timelocked data, I get this : >>> Warning: the trial definition in the configuration is inconsistent with >>> the >>> actual data >>>> >>>> In utilities/private/warning_once at 75 >>> >>>   In utilities/private/fixsampleinfo at 51 >>>   In ft_checkdata at 579 >>>   In ft_databrowser at 155 >>> Warning: failed to create sampleinfo field >>>> >>>> In utilities/private/warning_once at 75 >>> >>>   In utilities/private/fixsampleinfo at 83 >>>   In ft_checkdata at 579 >>>   In ft_databrowser at 155 >>>  So there must be something wrong right? >>> And yes I'm using the latest FT... >>> Many thanks, >>> --Marco >>> _______________________________________________ >>> fieldtrip mailing list >>> fieldtrip at donders.ru.nl >>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >>> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > -- > Jörn M. Horschig > PhD Student > Donders Institute for Brain, Cognition and Behaviour > Centre for Cognitive Neuroimaging > Radboud University Nijmegen > Neuronal Oscillations Group > > P.O. Box 9101 > NL-6500 HB Nijmegen > The Netherlands > > Contact: > E-Mail: jm.horschig at donders.ru.nl > Tel:    +31-(0)24-36-68493 > Web: http://www.ru.nl/donders > > Visiting address: > Trigon, room 2.30 > Kapittelweg 29 > NL-6525 EN Nijmegen > The Netherlands > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From eelke.spaak at donders.ru.nl Tue Aug 9 12:30:07 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Tue, 9 Aug 2011 12:30:07 +0200 Subject: [FieldTrip] databrowser and merged data In-Reply-To: References: <4E410A39.2060809@donders.ru.nl> Message-ID: PS: In other words, fixsampleinfo now generates an ad-hoc sampleinfo with no relation to the original dataset, if requested to do so. Marco's preprocessing steps are not relevant for this. 2011/8/9 Eelke Spaak : > Hi Jörn, > > As you can read in my email, I actually changed fixsampleinfo so that > it *does* reconstruct the sample info now :) > > I agree that sampleinfo does not make sense in the case of data that > is (1) timelock-averaged and (2) combined from multiple datasets, but > the issue here was to get the databrowser working on such data. Since > the data used internally by the databrowser (after the call to > ft_checkdata) is never exposed to the user, this seems like a good > solution to me. > > Best, > Eelke > > 2011/8/9 "Jörn M. Horschig" : >> Hi Marco and Eelke, >> >> sampleinfo cannot be reconstructed after resampling the data, because there >> is no way to precisely relate your resampled samples to the original >> samples. My guess would be that you, Marco, did that and therefore encounter >> the problem you mentioned. If that was so, I am not sure whether >> fixsampleinfo should actually recompute the sampleinfo field, because it >> makes absolutely no sense to have a sampleinfo field in your resampled data >> that relates back samples in the original data. >> Anyway, if you did not resample your data, then fixsampleinfo should work. >> Could you let us know how you preprocessed your data? >> >> Best, >> Jörn >> >> On 8/9/2011 11:00 AM, Eelke Spaak wrote: >>> >>> Hi Marco, >>> >>> The subfunction fixsampleinfo (used by ft_checkdata) attempts a number >>> of things in order to create a sampleinfo field. My guess is that it >>> detected a trial definition somewhere in your cfg "tree" (the data.cfg >>> and data.cfg.previous fields) that was inconsistent with your data, >>> and therefore refused to continue rebuilding a sampleinfo from >>> scratch. I have now changed this function so that it will always >>> reconstruct your sampleinfo, even if it detects an inconsistent trial >>> definition in a previous configuration structure. >>> >>> So, it should now work :) At least, with tomorrow's update on the FTP >>> server. >>> >>> Best, >>> Eelke >>> >>> 2011/8/8 Marco Dahmane: >>>> >>>> Hi Eelke, >>>> So in order :) >>>> 1) Fair enough. And I'm not using cfg.keeptrials='yes' anyway >>>> 2) I'm actually using databrowser because I find it really convenient to >>>> just remove and add channels, or see to which channel each line >>>> corresponds. >>>> I know I could do this using Matlab's plot, but I was just wondering why >>>> it >>>> wouldn't work with databrowser. >>>> 3) Ok this is weird then. Something that will probably help you, whenever >>>> I >>>> use databrowser on my timelocked data, I get this : >>>> Warning: the trial definition in the configuration is inconsistent with >>>> the >>>> actual data >>>>> >>>>> In utilities/private/warning_once at 75 >>>> >>>>   In utilities/private/fixsampleinfo at 51 >>>>   In ft_checkdata at 579 >>>>   In ft_databrowser at 155 >>>> Warning: failed to create sampleinfo field >>>>> >>>>> In utilities/private/warning_once at 75 >>>> >>>>   In utilities/private/fixsampleinfo at 83 >>>>   In ft_checkdata at 579 >>>>   In ft_databrowser at 155 >>>>  So there must be something wrong right? >>>> And yes I'm using the latest FT... >>>> Many thanks, >>>> --Marco >>>> _______________________________________________ >>>> fieldtrip mailing list >>>> fieldtrip at donders.ru.nl >>>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >>>> >>> _______________________________________________ >>> fieldtrip mailing list >>> fieldtrip at donders.ru.nl >>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> >> >> -- >> Jörn M. Horschig >> PhD Student >> Donders Institute for Brain, Cognition and Behaviour >> Centre for Cognitive Neuroimaging >> Radboud University Nijmegen >> Neuronal Oscillations Group >> >> P.O. Box 9101 >> NL-6500 HB Nijmegen >> The Netherlands >> >> Contact: >> E-Mail: jm.horschig at donders.ru.nl >> Tel:    +31-(0)24-36-68493 >> Web: http://www.ru.nl/donders >> >> Visiting address: >> Trigon, room 2.30 >> Kapittelweg 29 >> NL-6525 EN Nijmegen >> The Netherlands >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> > From jm.horschig at donders.ru.nl Tue Aug 9 14:11:34 2011 From: jm.horschig at donders.ru.nl (=?ISO-8859-1?Q?=22J=F6rn_M=2E_Horschig=22?=) Date: Tue, 09 Aug 2011 14:11:34 +0200 Subject: [FieldTrip] databrowser and merged data In-Reply-To: References: <4E410A39.2060809@donders.ru.nl> Message-ID: <4E4123F6.9000209@donders.ru.nl> Hi Eelke, I have read your mail and would like to emphasize that I disagree. If a sampleinfo field is constructed that cannot be used to infer anything from, it should not be done. Thus, the fix should not be in fixsampleinfo, because it will result in all kind of other issues when it comes to reconstructing sampleinfo, although it might help to solve this particular problem. Let's discuss this at the next FT meeting rather than here :) Best, Jörn On 8/9/2011 12:30 PM, Eelke Spaak wrote: > PS: In other words, fixsampleinfo now generates an ad-hoc sampleinfo > with no relation to the original dataset, if requested to do so. > Marco's preprocessing steps are not relevant for this. > > 2011/8/9 Eelke Spaak: >> Hi Jörn, >> >> As you can read in my email, I actually changed fixsampleinfo so that >> it *does* reconstruct the sample info now :) >> >> I agree that sampleinfo does not make sense in the case of data that >> is (1) timelock-averaged and (2) combined from multiple datasets, but >> the issue here was to get the databrowser working on such data. Since >> the data used internally by the databrowser (after the call to >> ft_checkdata) is never exposed to the user, this seems like a good >> solution to me. >> >> Best, >> Eelke >> >> 2011/8/9 "Jörn M. Horschig": >>> Hi Marco and Eelke, >>> >>> sampleinfo cannot be reconstructed after resampling the data, because there >>> is no way to precisely relate your resampled samples to the original >>> samples. My guess would be that you, Marco, did that and therefore encounter >>> the problem you mentioned. If that was so, I am not sure whether >>> fixsampleinfo should actually recompute the sampleinfo field, because it >>> makes absolutely no sense to have a sampleinfo field in your resampled data >>> that relates back samples in the original data. >>> Anyway, if you did not resample your data, then fixsampleinfo should work. >>> Could you let us know how you preprocessed your data? >>> >>> Best, >>> Jörn >>> >>> On 8/9/2011 11:00 AM, Eelke Spaak wrote: >>>> Hi Marco, >>>> >>>> The subfunction fixsampleinfo (used by ft_checkdata) attempts a number >>>> of things in order to create a sampleinfo field. My guess is that it >>>> detected a trial definition somewhere in your cfg "tree" (the data.cfg >>>> and data.cfg.previous fields) that was inconsistent with your data, >>>> and therefore refused to continue rebuilding a sampleinfo from >>>> scratch. I have now changed this function so that it will always >>>> reconstruct your sampleinfo, even if it detects an inconsistent trial >>>> definition in a previous configuration structure. >>>> >>>> So, it should now work :) At least, with tomorrow's update on the FTP >>>> server. >>>> >>>> Best, >>>> Eelke >>>> >>>> 2011/8/8 Marco Dahmane: >>>>> Hi Eelke, >>>>> So in order :) >>>>> 1) Fair enough. And I'm not using cfg.keeptrials='yes' anyway >>>>> 2) I'm actually using databrowser because I find it really convenient to >>>>> just remove and add channels, or see to which channel each line >>>>> corresponds. >>>>> I know I could do this using Matlab's plot, but I was just wondering why >>>>> it >>>>> wouldn't work with databrowser. >>>>> 3) Ok this is weird then. Something that will probably help you, whenever >>>>> I >>>>> use databrowser on my timelocked data, I get this : >>>>> Warning: the trial definition in the configuration is inconsistent with >>>>> the >>>>> actual data >>>>>> In utilities/private/warning_once at 75 >>>>> In utilities/private/fixsampleinfo at 51 >>>>> In ft_checkdata at 579 >>>>> In ft_databrowser at 155 >>>>> Warning: failed to create sampleinfo field >>>>>> In utilities/private/warning_once at 75 >>>>> In utilities/private/fixsampleinfo at 83 >>>>> In ft_checkdata at 579 >>>>> In ft_databrowser at 155 >>>>> So there must be something wrong right? >>>>> And yes I'm using the latest FT... >>>>> Many thanks, >>>>> --Marco >>>>> _______________________________________________ >>>>> fieldtrip mailing list >>>>> fieldtrip at donders.ru.nl >>>>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >>>>> >>>> _______________________________________________ >>>> fieldtrip mailing list >>>> fieldtrip at donders.ru.nl >>>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >>> >>> -- >>> Jörn M. Horschig >>> PhD Student >>> Donders Institute for Brain, Cognition and Behaviour >>> Centre for Cognitive Neuroimaging >>> Radboud University Nijmegen >>> Neuronal Oscillations Group >>> >>> P.O. Box 9101 >>> NL-6500 HB Nijmegen >>> The Netherlands >>> >>> Contact: >>> E-Mail: jm.horschig at donders.ru.nl >>> Tel: +31-(0)24-36-68493 >>> Web: http://www.ru.nl/donders >>> >>> Visiting address: >>> Trigon, room 2.30 >>> Kapittelweg 29 >>> NL-6525 EN Nijmegen >>> The Netherlands >>> >>> _______________________________________________ >>> fieldtrip mailing list >>> fieldtrip at donders.ru.nl >>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >>> > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip -- Jörn M. Horschig PhD Student Donders Institute for Brain, Cognition and Behaviour Centre for Cognitive Neuroimaging Radboud University Nijmegen Neuronal Oscillations Group P.O. Box 9101 NL-6500 HB Nijmegen The Netherlands Contact: E-Mail: jm.horschig at donders.ru.nl Tel: +31-(0)24-36-68493 Web: http://www.ru.nl/donders Visiting address: Trigon, room 2.30 Kapittelweg 29 NL-6525 EN Nijmegen The Netherlands From hubert.preissl at uni-tuebingen.de Wed Aug 10 11:16:08 2011 From: hubert.preissl at uni-tuebingen.de (Hubert Preissl) Date: Wed, 10 Aug 2011 11:16:08 +0200 Subject: [FieldTrip] =?windows-1252?q?Autumn_School_=93The_multimodal_brai?= =?windows-1252?q?n=94=3A_5th-6th_October_2011=2C_T=FCbingen?= In-Reply-To: <5B688DA3-CECE-42F5-9E0C-70F3FCB086FB@cmu.edu> References: <150C06A5-3B96-4F2C-B529-DB1079A36701@cmu.edu> <80BC061E-C9FD-46B5-B3D7-289260D35533@donders.ru.nl> <024FF5CF-ECE6-44ED-9C35-2D7636D797CF@cmu.edu> <6044F712-0B26-4637-B05B-9E89E9A7D206@donders.ru.nl> <5B688DA3-CECE-42F5-9E0C-70F3FCB086FB@cmu.edu> Message-ID: <4E424C58.6000406@uni-tuebingen.de> *Autumn School “The multimodal brain”: 5^th -6^th October 2011* Hello, we are pleased to announce the upcoming “Autumn school” in Tübingen organized by the MEG Center. For application and further information on invited speakers and scientific program please visit our website: http://www.mp.uni-tuebingen.de/mp/index.php?id=396 We look forward to meet you in Tübingen! Best regards , Hubert Preissl ps. sorry if you receive this mail several times based on some cross-posting. -- Dr. Hubert Preissl MEG Center phone: ++49-(0)7071-2987704 Otfried Müller Str 47 fax: ++49-(0)7071-295706 72076 Tübingen, Germany email: hubert.preissl at uni-tuebingen.de -------------- next part -------------- An HTML attachment was scrubbed... URL: From sylvana.schister at utah.edu Fri Aug 12 00:56:44 2011 From: sylvana.schister at utah.edu (Sylvana Schister) Date: Thu, 11 Aug 2011 16:56:44 -0600 Subject: [FieldTrip] Alignment of MRI and estimated sources Message-ID: <1313103404.26076.16.camel@sylvana-desktop> Hello Fieldtrippers! I am having trouble getting meaningful results after performing source analysis with DICS due to a misalignment of my MRI dataset. I am relatively new to FT and I will really appreciate your help. I am using the 'standard_BEM' files provided in the tutorials. I manage to realign the MRI to head coordinates using volumerealign, reslice and segment it. Then, I use the modified MRI to compute the forward model. The problem arises when proceeding to ft_sourceinterpolate and ft_sourceanalysis. As I understand, the 'mri' input to ft_sourceinterpolate must be the output of FT_READ_MRI or the filename of a MRI. The function crashes when using my realigned version of the data, therefore I am using the original file, 'standard_mri.mat'. This file contains no fiducials. No surprise the results are then messed up. How/ where do I need to specify the fiducials? Why can't I use the aligned mri? What is the correct way to do this? Thank you so much for your help! Sylvana From Santosh.VemaKrishnamurthy at nrc-cnrc.gc.ca Fri Aug 12 16:00:10 2011 From: Santosh.VemaKrishnamurthy at nrc-cnrc.gc.ca (Vema Krishnamurthy, Santosh) Date: Fri, 12 Aug 2011 07:00:10 -0700 Subject: [FieldTrip] Scalp segmentation from an MRI (.nii or .fif) Message-ID: Hello Everyone, I'm very new to Fieldtrip and I need to perform scalp segmentation from an mri file(.nii or .fif). I'm working on a project where I need the scalp surface for surface matching, so I need the scalp surface with high resolution which clearly shows the facial features (nose, eyes, lips, chin). Any help/suggestions relating to this are much appreciated. Thanks and Regards, --------------------------------------------------- Santosh Vema KrishnaMurthy, M.A.Sc Medical Devices Engineer NRC Institute for Biodiagnostics (Atlantic) Elekta/IWK MEG Research office Ph: (902) 470-3984 --------------------------------------------------- -------------- next part -------------- An HTML attachment was scrubbed... URL: From Gregor.Volberg at psychologie.uni-regensburg.de Fri Aug 12 18:47:30 2011 From: Gregor.Volberg at psychologie.uni-regensburg.de (Gregor Volberg) Date: Fri, 12 Aug 2011 18:47:30 +0200 Subject: [FieldTrip] Antw: Scalp segmentation from an MRI (.nii or .fif) Message-ID: <4E457543020000570000A631@gwsmtp1.uni-regensburg.de> Dear Santosh, for that purpose you might want to have a look at the NFT toolbox from SCCN, http://sccn.ucsd.edu/nft/ . It does high resolution segmentations for brain, csf, scalp and skin surfaces. If you need the segmentation for computing forward solutions with BEM head models, then I would discourage a too fine resolution, though. It strongly increases the computational load. Best regards, Gregor -- Dr. rer. nat. Gregor Volberg ( mailto:gregor.volberg at psychologie.uni-regensburg.de ) University of Regensburg Institute for Experimental Psychology 93040 Regensburg, Germany Tel: +49 941 943 3862 Fax: +49 941 943 3233 http://www.psychologie.uni-regensburg.de/Greenlee/team/volberg/volberg.html >>> "Vema Krishnamurthy, Santosh" 12.08.11 16.42 Uhr >>> Hello Everyone, I'm very new to Fieldtrip and I need to perform scalp segmentation from an mri file(.nii or .fif). I'm working on a project where I need the scalp surface for surface matching, so I need the scalp surface with high resolution which clearly shows the facial features (nose, eyes, lips, chin). Any help/suggestions relating to this are much appreciated. Thanks and Regards, --------------------------------------------------- Santosh Vema KrishnaMurthy, M.A.Sc Medical Devices Engineer NRC Institute for Biodiagnostics (Atlantic) Elekta/IWK MEG Research office Ph: (902) 470-3984 --------------------------------------------------- -------------- next part -------------- An HTML attachment was scrubbed... URL: From jan.schoffelen at donders.ru.nl Sat Aug 13 18:23:49 2011 From: jan.schoffelen at donders.ru.nl (jan-mathijs schoffelen) Date: Sat, 13 Aug 2011 18:23:49 +0200 Subject: [FieldTrip] Fwd: ft_definetrial crash References: Message-ID: <4B01B4FC-FF35-435F-8167-F2A82C0D1B9E@donders.ru.nl> Hi Tobias, I forward your question to the discussion list, so that many may benefit from the discussion. I am not sure what is going on and I don't know whether other people have encountered this. Are you calling the function from the correct directory? In other words, you should not be within the *.ds directory itself. You could also try to explicitly define cfg.datafile and cfg.headerfile (and omit the cfg.dataset). Let me know when the problem persists. Best wishes, Jan-Mathijs Begin forwarded message: > From: Tobias Overath > Date: August 3, 2011 12:37:15 PM GMT+02:00 > To: jan.schoffelen at donders.ru.nl > Subject: ft_definetrial crash > > Hi Jan-Mathijs, > > I participated in the MEG FieldTrip course this past April and have finally managed to get my first MEG dataset from the CTF machine here at the FIL. However, I am already crashing at the first step, i.e. when running ft_definetrial. > > This is my code: > cfg = []; > cfg.dataset = 'S01_01.ds'; > cfg.trialdef.eventtype = 'frontpanel trigger'; > cfg.trialdef.eventvalue = 1; > cfg.trialdef.prestim = .25; > cfg.trialdef.poststim = 2.75; > cfg = ft_definetrial(cfg); > > It recognises correctly that it's supposed to read the header from S01_01.ds/S01_01.res, but it then tells me that it can't find the file S01_01.meg4 (which is true, because it should be S01_01.ds/S01_01.meg4). I presume this is a pretty common error, but it occurs with two FieldTrip versions that I have downloaded (very recent ones). I am probably missing something very obvious... > > Incidentally, is there no other way to search the FieldTrip email archives than to download a zipped text file for each month of the previous 7 years?!? > > Thanks, > > Tobias > > > > -- > Tobias Overath, PhD > UCL Ear Institute > University College London > 332 Grays Inn Road > London, WC1X 8EE > UK > http://www.homepages.ucl.ac.uk/~skgtjto > > > > > Dr. J.M. (Jan-Mathijs) Schoffelen Donders Institute for Brain, Cognition and Behaviour, Centre for Cognitive Neuroimaging, Radboud University Nijmegen, The Netherlands J.Schoffelen at donders.ru.nl Telephone: +31-24-3614793 -------------- next part -------------- An HTML attachment was scrubbed... URL: From r.oostenveld at donders.ru.nl Mon Aug 15 10:10:06 2011 From: r.oostenveld at donders.ru.nl (Robert Oostenveld) Date: Mon, 15 Aug 2011 10:10:06 +0200 Subject: [FieldTrip] Invitation to contribute to a Special Issue on Brain Oscillations during Language Processing References: Message-ID: <021E015B-2CB1-46FA-96E3-0DA3922CEB58@donders.ru.nl> Begin forwarded message: > Dear colleagues, > We would like to invite you to contribute your research to our special issue on the role of brain oscillations in language processing, to appear in Frontiers in Language Science. > You can visit the web site at: > http://www.frontiersin.org/languagesciences/specialtopics/brain_oscillations_during_lang/397 > See detailed description below. > The call has been very successful so far and prominent figures in the field have joined us in this project. > We are looking forward to receiving your research. > Best wishes, > Lucia Melloni & Marcela Pena > > > Brain Oscillations during Language Processing: from Perception to Production > > Deadline for abstract submission: 01 Sep 2011 > Deadline for full article submission: 15 Dec 2011 > > Language processing is a seemingly effortless task that requires the integration of speech units (e.g., phonemes, syllables, words, etc.) occurring at different rates. In particular, temporal binding for speech should occur within and across different temporal scales, necessitating multiple simultaneous windows of integration for prosodic, semantic, syntactic and pragmatic processing. Recent evidence suggests that neuronal oscillations may reflect both tracking linguistic units at their individual rhythms as well as integrating speech units over a large range of temporal scales. > > The present Research Topic would like to evaluate current theories and evidence for a mechanistic role of neuronal oscillations in measuring language processing, covering the latest advances brought about by EEG, MEG and fMRI imaging methods. Our main focus is to highlight innovative and foundational studies that go beyond methodological issues and advance our theoretical understanding of the role of brain oscillations in language processing. Contributions from the pioneers of this field are selected, illustrating how the study of brain oscillations has allowed investigating theoretically relevant questions that could not be addressed by more traditional methods. The topic thus aims at deepening our mechanistic understanding of language processing and bringing us closer to bridging the gap between brain, mind and behavior for the crucial cognitive function of speech. > > Hosted By: > Marcela Pena, Catholic University of Chile, Chile > Lucia Melloni, Max Planck Institute for Brain Research, Germany -------------- next part -------------- An HTML attachment was scrubbed... URL: From Santosh.VemaKrishnamurthy at nrc-cnrc.gc.ca Mon Aug 15 14:06:43 2011 From: Santosh.VemaKrishnamurthy at nrc-cnrc.gc.ca (Vema Krishnamurthy, Santosh) Date: Mon, 15 Aug 2011 08:06:43 -0400 Subject: [FieldTrip] fieldtrip Digest, Vol 9, Issue 14 In-Reply-To: References: Message-ID: Hello Gregor, Thanks for the suggestions, I will give this a try. I'm not doing forward modeling with the high resolution surfaces. I'm trying to perform surface matching technique with scalp surface and the digitization data to do the co-registration of MEG and MRI. Thanks and Regards, --------------------------------------------------- Santosh Vema KrishnaMurthy, M.A.Sc Medical Devices Engineer NRC Institute for Biodiagnostics (Atlantic) Elekta/IWK MEG Research office Ph: (902) 470-3984 --------------------------------------------------- ________________________________________ From: fieldtrip-bounces at donders.ru.nl [fieldtrip-bounces at donders.ru.nl] On Behalf Of fieldtrip-request at donders.ru.nl [fieldtrip-request at donders.ru.nl] Sent: Saturday, August 13, 2011 7:00 AM To: fieldtrip at donders.ru.nl Subject: fieldtrip Digest, Vol 9, Issue 14 Send fieldtrip mailing list submissions to fieldtrip at donders.ru.nl To subscribe or unsubscribe via the World Wide Web, visit http://mailman.science.ru.nl/mailman/listinfo/fieldtrip or, via email, send a message with subject or body 'help' to fieldtrip-request at donders.ru.nl You can reach the person managing the list at fieldtrip-owner at donders.ru.nl When replying, please edit your Subject line so it is more specific than "Re: Contents of fieldtrip digest..." Today's Topics: 1. Scalp segmentation from an MRI (.nii or .fif) (Vema Krishnamurthy, Santosh) 2. Antw: Scalp segmentation from an MRI (.nii or .fif) (Gregor Volberg) ---------------------------------------------------------------------- Message: 1 Date: Fri, 12 Aug 2011 07:00:10 -0700 From: "Vema Krishnamurthy, Santosh" To: "fieldtrip at donders.ru.nl" Subject: [FieldTrip] Scalp segmentation from an MRI (.nii or .fif) Message-ID: Content-Type: text/plain; charset="iso-8859-1" Hello Everyone, I'm very new to Fieldtrip and I need to perform scalp segmentation from an mri file(.nii or .fif). I'm working on a project where I need the scalp surface for surface matching, so I need the scalp surface with high resolution which clearly shows the facial features (nose, eyes, lips, chin). Any help/suggestions relating to this are much appreciated. Thanks and Regards, --------------------------------------------------- Santosh Vema KrishnaMurthy, M.A.Sc Medical Devices Engineer NRC Institute for Biodiagnostics (Atlantic) Elekta/IWK MEG Research office Ph: (902) 470-3984 --------------------------------------------------- -------------- next part -------------- An HTML attachment was scrubbed... URL: ------------------------------ Message: 2 Date: Fri, 12 Aug 2011 18:47:30 +0200 From: "Gregor Volberg" To: Subject: [FieldTrip] Antw: Scalp segmentation from an MRI (.nii or .fif) Message-ID: <4E457543020000570000A631 at gwsmtp1.uni-regensburg.de> Content-Type: text/plain; charset="us-ascii" Dear Santosh, for that purpose you might want to have a look at the NFT toolbox from SCCN, http://sccn.ucsd.edu/nft/ . It does high resolution segmentations for brain, csf, scalp and skin surfaces. If you need the segmentation for computing forward solutions with BEM head models, then I would discourage a too fine resolution, though. It strongly increases the computational load. Best regards, Gregor -- Dr. rer. nat. Gregor Volberg ( mailto:gregor.volberg at psychologie.uni-regensburg.de ) University of Regensburg Institute for Experimental Psychology 93040 Regensburg, Germany Tel: +49 941 943 3862 Fax: +49 941 943 3233 http://www.psychologie.uni-regensburg.de/Greenlee/team/volberg/volberg.html >>> "Vema Krishnamurthy, Santosh" 12.08.11 16.42 Uhr >>> Hello Everyone, I'm very new to Fieldtrip and I need to perform scalp segmentation from an mri file(.nii or .fif). I'm working on a project where I need the scalp surface for surface matching, so I need the scalp surface with high resolution which clearly shows the facial features (nose, eyes, lips, chin). Any help/suggestions relating to this are much appreciated. Thanks and Regards, --------------------------------------------------- Santosh Vema KrishnaMurthy, M.A.Sc Medical Devices Engineer NRC Institute for Biodiagnostics (Atlantic) Elekta/IWK MEG Research office Ph: (902) 470-3984 --------------------------------------------------- -------------- next part -------------- An HTML attachment was scrubbed... URL: ------------------------------ _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip End of fieldtrip Digest, Vol 9, Issue 14 **************************************** From shreesb at yahoo.com Mon Aug 15 21:30:22 2011 From: shreesb at yahoo.com (shree b) Date: Mon, 15 Aug 2011 12:30:22 -0700 (PDT) Subject: [FieldTrip] Header file for TDT system Message-ID: <1313436622.23112.YahooMailClassic@web162012.mail.bf1.yahoo.com> Hello, I am new to field-trip and wanted to use your analysis on 16 channel continous LFP recordings from the TDT (Tucker Davis Technology) system. As there isn't a provision to read TDT header files I tried to input the values directly into the structure array. 1)I have already imported my data into Matlab and have the information that is need to be included in cfg.headerfile in a structure array: hdr=[]; hdr.Fs=1000; hdr.nChans=16; hdr.nSamples= abfi.dataPtsPerChan; %abfi.dataPtsPerChan=620646 hdr.nSamplesPre=0; hdr.nTrials=1; hdr.label= abfi.recChNames; %abfi.recChNames is a cell array with the channel names: %IN0,IN1,IN2,IN3,IN4,IN5,IN6,IN7,IN8,IN9,IN10,IN11,IN12,IN13,IN14,IN15 2) The data is in a mat file with all 16 channels labeled as before 3) I defined trl as: a=1; b=length(IN0); c=0; trl=[a b c];% single trial no offset cfg=struct('headerfile',hdr,'datafile', data,'trl',trl) This approach isn't working, could you please tell me how I should specify the values for the header file and data structure. thanks very much Shilpa From kirihara-tky at umin.ac.jp Tue Aug 16 06:00:11 2011 From: kirihara-tky at umin.ac.jp (Kenji Kirihara) Date: Tue, 16 Aug 2011 06:00:11 JST Subject: [FieldTrip] Time-frequency analysis Message-ID: <08160600.ME2066201@umin.org> Dear FieldTrip mailing list members, I am interested in time-frequency analysis of EEG and have a question about ft_freqanalysis. I applied ft_freqanalysis to my data for time-frequency analysis using wavelets. Power at each time-requency point is calculated. However, power at each point seems to be too large. For example, power at 4 Hz, 0 ms, O2 is 1218, power at 10 Hz is 730, and power at 20 Hz is 94. These values seem to be too large compared to power calculated using FFT for frequency analysis. I would be grateful if someone tells me why this happens. Sincerely, Kenji Kirihara From BelluscB at ninds.nih.gov Mon Aug 15 23:25:21 2011 From: BelluscB at ninds.nih.gov (Belluscio, Beth (NIH/NINDS) [E]) Date: Mon, 15 Aug 2011 17:25:21 -0400 Subject: [FieldTrip] preparing the forward model in beamforming Message-ID: <794DFDD3C128EF44AC49ADC58FD0090C057456B65B@NIHMLBX10.nih.gov> Hi Fieldtrip users- I am trying to learn to use the beamforming technique as outlined in the tutorial. I have a dicom MRI that I used to create a file that contains markers for the fiducials. I then used this in ft_read_mri and ft_volumesegment - both of which seemed to work fine. Then, when I used ft_prepare_singleshell, I get this error message: "??? Undefined function or method 'imfill' for input arguments of type 'char'. Error in ==> prepare_mesh_segmentation at 68 Seg = imfill((mri.seg==cfg.tisse(i)), 'holes')" I am used Fieldtrip 20110721 Any suggestions? Beth. Beth Belluscio MD-PhD Clinical Fellow Human Motor Control Section NINDS, NIH 301-402-3495 -------------- next part -------------- An HTML attachment was scrubbed... URL: From Don.Rojas at ucdenver.edu Mon Aug 15 23:57:07 2011 From: Don.Rojas at ucdenver.edu (Rojas, Don) Date: Mon, 15 Aug 2011 15:57:07 -0600 Subject: [FieldTrip] preparing the forward model in beamforming In-Reply-To: <794DFDD3C128EF44AC49ADC58FD0090C057456B65B@NIHMLBX10.nih.gov> References: <794DFDD3C128EF44AC49ADC58FD0090C057456B65B@NIHMLBX10.nih.gov> Message-ID: <2D825D9F-FEDA-4B1C-B401-38C3878B4E36@ucdenver.edu> Beth, The matlab function imfill.m appears to be missing for your matlab install. It is part of the matlab image processing toolbox, which is not part of the base matlab package. You could check with your local matlab admin to see if you should have the image processing toolbox. FieldTrip should probably check for the existence of this toolbox when used so that it can throw a more meaningful error message. Best, Don On Aug 15, 2011, at 3:25 PM, Belluscio, Beth (NIH/NINDS) [E] wrote: Hi Fieldtrip users- I am trying to learn to use the beamforming technique as outlined in the tutorial. I have a dicom MRI that I used to create a file that contains markers for the fiducials. I then used this in ft_read_mri and ft_volumesegment – both of which seemed to work fine. Then, when I used ft_prepare_singleshell, I get this error message: “??? Undefined function or method ‘imfill’ for input arguments of type ‘char’. Error in ==> prepare_mesh_segmentation at 68 Seg = imfill((mri.seg==cfg.tisse(i)), ‘holes’)” I am used Fieldtrip 20110721 Any suggestions? Beth. Beth Belluscio MD-PhD Clinical Fellow Human Motor Control Section NINDS, NIH 301-402-3495 _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip ----------------------- Don Rojas, Ph.D. Associate Professor of Psychiatry U. of Colorado Denver Anschutz Medical Campus Director, UCD Magnetoencephalography Lab 13001 E. 17th Pl F546 Aurora, CO 80045 303-724-4994 -------------- next part -------------- An HTML attachment was scrubbed... URL: From jan.schoffelen at donders.ru.nl Tue Aug 16 10:01:28 2011 From: jan.schoffelen at donders.ru.nl (jan-mathijs schoffelen) Date: Tue, 16 Aug 2011 10:01:28 +0200 Subject: [FieldTrip] Time-frequency analysis In-Reply-To: <08160600.ME2066201@umin.org> References: <08160600.ME2066201@umin.org> Message-ID: <5FE3FAAE-6A78-4033-B84C-AEB55E844285@donders.ru.nl> Dear Kenji, In addition of applying the fft on the data, fieldtrip normalises the fft with a number that equals 2/Nsmp, where Nsmp is the length of the wavelet. Best wishes, Jan-Mathijs On Aug 15, 2011, at 11:00 PM, Kenji Kirihara wrote: > Dear FieldTrip mailing list members, > > I am interested in time-frequency analysis of EEG and have > a question about ft_freqanalysis. I applied ft_freqanalysis to > my data for time-frequency analysis using wavelets. Power at each > time-requency point is calculated. However, power at each point > seems to be too large. For example, power at 4 Hz, 0 ms, O2 is 1218, > power at 10 Hz is 730, and power at 20 Hz is 94. These values seem > to be too large compared to power calculated using FFT for frequency > analysis. I would be grateful if someone tells me why this happens. > > Sincerely, > > Kenji Kirihara > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip Dr. J.M. (Jan-Mathijs) Schoffelen Donders Institute for Brain, Cognition and Behaviour, Centre for Cognitive Neuroimaging, Radboud University Nijmegen, The Netherlands J.Schoffelen at donders.ru.nl Telephone: +31-24-3614793 From eelke.spaak at donders.ru.nl Tue Aug 16 10:36:16 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Tue, 16 Aug 2011 10:36:16 +0200 Subject: [FieldTrip] Header file for TDT system In-Reply-To: <1313436622.23112.YahooMailClassic@web162012.mail.bf1.yahoo.com> References: <1313436622.23112.YahooMailClassic@web162012.mail.bf1.yahoo.com> Message-ID: Dear Shilpa, If you already have the data and related information in Matlab, it is probably best to skip the entire trial-definition step when converting your data to FieldTrip format. It's quite straightforward to convert data into a format that FT can handle, especially if you just want it to have a single segment (or 'trial'). In your case, from the top of my head, you could initialize a structure with the following fields: data.label = abfi.recChNames; data.trial = {yourDataMatrix}; % assuming size(yourDataMatrix) == [nChans nSamples] data.fsample = 1000; data.time = {0:1/data.fsample:lastTimePoint}; data.sampleinfo = [1 nSamples]; This is just about the bare minimum that FT requires, which in your case should be sufficient. Note that both the data.trial and data.time fields should be cell arrays, with one element per trial in your data. For more detailed information about how FT raw data should look, see the following entry on our wiki: http://fieldtrip.fcdonders.nl/faq/how_are_the_various_data_structures_defined . Hope this was of help, Best, Eelke 2011/8/15 shree b : > Hello, > I am new to field-trip and wanted to use your analysis on 16 channel continous LFP recordings from the TDT (Tucker Davis Technology) system. As there isn't a provision to read TDT header files I tried to input the values directly into the structure array. > > 1)I have already imported my data into Matlab and have the information that is need to be included in cfg.headerfile in a structure array: > > hdr=[]; > > hdr.Fs=1000; > hdr.nChans=16; > hdr.nSamples= abfi.dataPtsPerChan; %abfi.dataPtsPerChan=620646 > hdr.nSamplesPre=0; > hdr.nTrials=1; > hdr.label= abfi.recChNames; > %abfi.recChNames is a cell array with the channel names:    %IN0,IN1,IN2,IN3,IN4,IN5,IN6,IN7,IN8,IN9,IN10,IN11,IN12,IN13,IN14,IN15 > > 2) The data is in a mat file with all 16 channels labeled as before > > 3) I defined trl as: > a=1; > b=length(IN0); > c=0; > > trl=[a b c];% single trial no offset > > cfg=struct('headerfile',hdr,'datafile', data,'trl',trl) > > This approach isn't working, could you please tell me how I should specify the values for the header file and data structure. > > thanks very much > Shilpa > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From jakobala at hotmail.com Tue Aug 16 14:09:05 2011 From: jakobala at hotmail.com (jakob kaiser) Date: Tue, 16 Aug 2011 14:09:05 +0200 Subject: [FieldTrip] scaling frequency data for comparison Message-ID: Dear list,I wanted to compare the outputs from different ft_freqanalysis-methods to see if they make significant differences in the final test results of my data. This were the parameters used: cfg.output = 'pow';cfg.channel = 'all';cfg.method = 'mtmconvol';cfg.taper = 'hanning';cfg.foi = 6:2:20; cfg.t_ftimwin = ones(length(cfg.foi),1).*0.5; cfg.toi = -0.5:0.05:2; WaveletConfig.method = 'wavelet'; WaveletConfig.output = 'pow'; WaveletConfig.foi = 6:2:20; WaveletConfig.toi = -0.5:0.05:2; (I chose these two methods because I found them used un papers relevant for my work)Now, as has been discussed on the list before, the scaling of the output data is very different. While the mtconvol-hanning-method gives me data points between roughly -2 an 1, with the wavelet-method I have data between roughly -800 and 600. From the (btw very helpful) tutorials and the previous list discussions I understand that this is a normal byproduct of the mathematical methods employed. However, for comparing different outputs it would be useful to scale both produced data sets to the same unit, so to speak. As both methods are supposed to measure the same thing, it should someone be possible to scale them to similar sizes, shouldn't it? Is there such a way? (Ultimately, I will have to compare the data results to results from a completely different program, which seem to have a third, different scaling altogether, so I am really puzzled how to handle the differences in scale)Generally, although this has been mentioned before, I am still not sure, if it is possible to put a certain unit to the results. Power is often reported in mV^2/Hz. If my initial data is in mV, would it be correct to say, that the hanning-method describe above delivers data in mV^2/Hz? Would this be correct for the Wavelet-method? Obviously, both cannot be true, because the scaling is extremely different. So, is it possible to specify a unit for any of these methods? Is there a way to convert the output of ft_freqanalysis-methods to this or another meaningful unit?I would be very grateful, if someone could give me a hint, how to interpret the data points here.Thank you for reading + thanks to the programmers for fieldtrip, which I like a lot Jakob -------------- next part -------------- An HTML attachment was scrubbed... URL: From eelke.spaak at donders.ru.nl Tue Aug 16 14:20:45 2011 From: eelke.spaak at donders.ru.nl (Eelke Spaak) Date: Tue, 16 Aug 2011 14:20:45 +0200 Subject: [FieldTrip] scaling frequency data for comparison In-Reply-To: References: Message-ID: Dear Jakob, I can't give you any definitive answer on the power units of the different freqanalysis algorithms, but you might find the function ft_freqbaseline helpful. Absolute power values are usually not very meaningful in neuroscience research, since they depend on a lot of non-brain factors. Ft_freqbaseline performs a baseline correction (either relative, absolute, or percentage) on your power values. Across different spectral estimation methods, the relative power change should remain the same (since it is unitless). Best, Eelke 2011/8/16 jakob kaiser : > Dear list, > I wanted to compare the outputs from different ft_freqanalysis-methods to > see if they make significant differences in the final test results of my > data. This were the parameters used: > > cfg.output       = 'pow'; > cfg.channel      = 'all'; > cfg.method       = 'mtmconvol'; > cfg.taper        = 'hanning'; > cfg.foi          = 6:2:20; > cfg.t_ftimwin    = ones(length(cfg.foi),1).*0.5; > cfg.toi          = -0.5:0.05:2; > > WaveletConfig.method = 'wavelet'; > WaveletConfig.output = 'pow'; > WaveletConfig.foi = 6:2:20; > WaveletConfig.toi = -0.5:0.05:2; > (I chose these two methods because I found them used un papers relevant for > my work) > Now, as has been discussed on the list before, the scaling of the output > data is very different. While the mtconvol-hanning-method gives me data > points between roughly -2 an 1, with the wavelet-method I have data between > roughly -800 and 600. From the (btw very helpful) tutorials and the previous > list discussions I understand that this is a normal byproduct of > the mathematical methods employed. However, for comparing different outputs > it would be useful to scale both produced data sets to the same unit, so to > speak. As both methods are supposed to measure the same thing, it should > someone be possible to scale them to similar sizes, shouldn't it? Is there > such a way? (Ultimately, I will have to compare the data results to results > from a completely different program, which seem to have a third, different > scaling altogether, so I am really puzzled how to handle the differences in > scale) > Generally, although this has been mentioned before, I am still not sure, if > it is possible to put a certain unit to the results. Power is often reported > in mV^2/Hz. If my initial data is in mV, would it be correct to say, that > the hanning-method describe above delivers data in mV^2/Hz? Would this be > correct for the Wavelet-method? Obviously, both cannot be true, because the > scaling is extremely different. So, is it possible to specify a unit for any > of these methods? Is there a way to convert the output > of ft_freqanalysis-methods to this or another meaningful unit? > I would be very grateful, if someone could give me a hint, how to interpret > the data points here. > Thank you for reading + thanks to the programmers for fieldtrip, which I > like a lot > Jakob > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From kirihara-tky at umin.ac.jp Wed Aug 17 02:33:15 2011 From: kirihara-tky at umin.ac.jp (Kenji Kirihara) Date: Wed, 17 Aug 2011 02:33:15 JST Subject: [FieldTrip] Time-frequency analysis Message-ID: <08170233.ME2758201@umin.org> Dear Jan-Mathijs, Thank you for your advice. Sincerely, Kenji, In message <5FE3FAAE-6A78-4033-B84C-AEB55E844285 at donders.ru.nl> fieldtrip at donder s.ru.nl (jan-mathijs schoffelen) wrote: > From: jan-mathijs schoffelen > Date: Tue, 16 Aug 2011 10:01:28 +0200 > To: Email discussion list for the FieldTrip project > Subject: Re: [FieldTrip] Time-frequency analysis > > Dear Kenji, > > In addition of applying the fft on the data, fieldtrip normalises the fft with a number that equals 2/Nsmp, where Nsmp is the length of the wavelet. > > Best wishes, > > Jan-Mathijs > > > On Aug 15, 2011, at 11:00 PM, Kenji Kirihara wrote: > > > Dear FieldTrip mailing list members, > > > > I am interested in time-frequency analysis of EEG and have > > a question about ft_freqanalysis. I applied ft_freqanalysis to > > my data for time-frequency analysis using wavelets. Power at each > > time-requency point is calculated. However, power at each point > > seems to be too large. For example, power at 4 Hz, 0 ms, O2 is 1218, > > power at 10 Hz is 730, and power at 20 Hz is 94. These values seem > > to be too large compared to power calculated using FFT for frequency > > analysis. I would be grateful if someone tells me why this happens. > > > > Sincerely, > > > > Kenji Kirihara > > _______________________________________________ > > fieldtrip mailing list > > fieldtrip at donders.ru.nl > > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > Dr. J.M. (Jan-Mathijs) Schoffelen > Donders Institute for Brain, Cognition and Behaviour, > Centre for Cognitive Neuroimaging, > Radboud University Nijmegen, The Netherlands > J.Schoffelen at donders.ru.nl > Telephone: +31-24-3614793 > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From g.piantoni at nin.knaw.nl Wed Aug 17 12:43:23 2011 From: g.piantoni at nin.knaw.nl (Gio Piantoni) Date: Wed, 17 Aug 2011 12:43:23 +0200 Subject: [FieldTrip] Granger convention Message-ID: Hi all, The connectivity toolbox is very powerful and straightforward to use, thanks for that! However, I did not find any information about the convention used by Fieldtrip to represent asymmetrical connectivity measures (such as granger causality), f.e. in the help of the functions, in http://fieldtrip.fcdonders.nl/faq/in_what_way_can_frequency_domain_data_be_represented_in_fieldtrip nor in http://fieldtrip.fcdonders.nl/development/connectivity_tutorial Do the values in data.grangerspctrm(1, 2, :) represent the granger-causality values from channel 1 to channel 2 or the other way round? BTW, is it possible to run time-domain granger causality in Fieldtrip? Thanks! Gio -- Giovanni Piantoni, Ph.D. student Dept. Sleep & Cognition Netherlands Institute for Neuroscience Meibergdreef 47 1105 BA Amsterdam (NL) +31 (0)20 5665492 g.piantoni at nin.knaw.nl www.giovannipiantoni.com From jpnv2006 at gmail.com Wed Aug 17 14:58:07 2011 From: jpnv2006 at gmail.com (Juan Pablo Neira) Date: Wed, 17 Aug 2011 14:58:07 +0200 Subject: [FieldTrip] Forward solution using concentric spheres and BEM models Message-ID: Hello, I am working with an individual MRI and EEG data to do the forward solution with two different volume's model (concentric spheres and BEM), so at the end i can compare the results.  But i did not have the expected results of the concentric spheres model until now. This is my script: 1. Read individual MRI mri=ft_read_mri('data_patient_1\******.hdr'); %I got this "Warning: flipping 1st dimension (L-R) to obtain volume in neurological convention" %How can i do a homogenous transformation matrix, using the voxel dimensions %that are specified in hdr.dime.pixdim? 2. Realign to 'head coordinates' 3. MRI Reslicing 4. MRI segmentation: (brain, skull,scalp) 5. Create geometrical description of the brain, skull and scalp (3 compartments) cfg = []; cfg.interactive = 'no'; %segmentation method cfg.numvertices = 3000; cfg.sourceunits = 'mm'; cfg.downsample = 2; cfg.numcompartments = 3; cfg.tissue = {'brain' 'skull' 'scalp'}; cfg.smooth = 'yes'; bnd = ft_prepare_mesh(cfg, mri_segment); %The output just give me the geometrical description of the brain. How can i get the geometrical description of %the skull and scalp also to proceed to create a 3 spheres concentric sphere model? 6. If i have the 3 geometrical descriptions (brain, skull and scalp). Create a 3 concentric spheres model. cfg = []; cfg.headshape = [bnd1 bnd2 bnd3]; % brain- skull - scalp cfg.conductivity = [0.33 0.0042 0.331]; [vol, cfg] = prepare_concentricspheres(cfg); %I got the model but part of the brain is outside of its sphere and also the skull. How can I increase the radius so % the whole brain will be inside its sphere and also the skull? The sphere of the scalp fix good. I hope someone can help me solve these questions. Regards, Juan Pablo From sysoevao at psychiatry.wustl.edu Thu Aug 18 05:39:11 2011 From: sysoevao at psychiatry.wustl.edu (Sysoeva, Olga Vladimirovna) Date: Wed, 17 Aug 2011 22:39:11 -0500 Subject: [FieldTrip] mixed between/within subject Anova Message-ID: <76DC57362137854CB9592B496182D5FA0506E253FF@wusmexmbx2.medpriv.wucon.wustl.edu> Dear fieldtrip list members, I'd like to use mixed between subjects (3 groups) and within subject (2 factors each has 2 levels) Anova design to compare the ERPs using cluster permutation statistics (ft_timelockananlysis). But to my understanding it is impossible. Am I right? You comments are highly appreciated, Best Regards, Olga. Olga Sysoeva, Research Associate, PhD Washington University School of Medicine Campus Box 8134 660 South Euclid Ave Saint Louis, MO 63110-9909 The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. From sysoevao at psychiatry.wustl.edu Thu Aug 18 05:41:36 2011 From: sysoevao at psychiatry.wustl.edu (Sysoeva, Olga Vladimirovna) Date: Wed, 17 Aug 2011 22:41:36 -0500 Subject: [FieldTrip] cluster analysis for peak/latency data Message-ID: <76DC57362137854CB9592B496182D5FA0506E253FD@wusmexmbx2.medpriv.wucon.wustl.edu> Dear fieldtrip list members, I’m applying cluster permutation algorithm, implicated in Fieldtrip software to my EEG data. But I’m using it in a little bit different way. I want to cluster my data from 32 channels for peak amplitude and latency measures, not raw ERP data. So, as an input I have only chn*subjects matrix, without time dimension. Technically the script works fine (ft_timelockstatistics), and, from my point of view, it also makes perfect theoretic sense: we can examined both peak and latency of the component ( in point by point comparison it might be confusing to disentangle these effects). Certainly, the main motivation for this cluster-permutation test was to deal with multiple comparison problem and to conduct exploratory investigation (huge datasets). But in case of looking a specific component cluster permutation analysis is helping to deal with MCP introduced by multiple channels. Am I right? You comments are highly appreciated, Best Regards, Olga. Olga Sysoeva, Research Associate, PhD Washington University School of Medicine Campus Box 8134 660 South Euclid Ave Saint Louis, MO 63110-9909 The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. From e.maris at donders.ru.nl Thu Aug 18 07:40:32 2011 From: e.maris at donders.ru.nl (Eric Maris) Date: Thu, 18 Aug 2011 07:40:32 +0200 Subject: [FieldTrip] cluster analysis for peak/latency data In-Reply-To: <76DC57362137854CB9592B496182D5FA0506E253FD@wusmexmbx2.medpriv.wucon.wustl.edu> References: <76DC57362137854CB9592B496182D5FA0506E253FD@wusmexmbx2.medpriv.wucon.wustl.edu> Message-ID: <024f01cc5d69$58e406f0$0aac14d0$@maris@donders.ru.nl> Dear Olga, > But I'm using it in a little bit different way. I want to cluster my data from 32 > channels for peak amplitude and latency measures, not raw ERP data. So, as > an input I have only chn*subjects matrix, without time dimension. > Technically the script works fine (ft_timelockstatistics), and, from my point > of view, it also makes perfect theoretic sense: we can examined both peak > and latency of the component ( in point by point comparison it might be > confusing to disentangle these effects). Certainly, the main motivation for > this cluster-permutation test was to deal with multiple comparison problem > and to conduct exploratory investigation (huge datasets). But in case of > looking a specific component cluster permutation analysis is helping to deal > with MCP introduced by multiple channels. Am I right? If you do this separately for peak latency and peak amplitude, I see no problem with this approach. Best, Eric > > You comments are highly appreciated, > > Best Regards, > Olga. > > > > Olga Sysoeva, > Research Associate, PhD > Washington University School of Medicine > Campus Box 8134 > 660 South Euclid Ave > Saint Louis, MO 63110-9909 > > The materials in this message are private and may contain Protected > Healthcare Information or other information of a sensitive nature. If you are > not the intended recipient, be advised that any unauthorized use, disclosure, > copying or the taking of any action in reliance on the contents of this > information is strictly prohibited. If you have received this email in error, > please immediately notify the sender via telephone or return mail. > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From e.maris at donders.ru.nl Thu Aug 18 07:50:08 2011 From: e.maris at donders.ru.nl (Eric Maris) Date: Thu, 18 Aug 2011 07:50:08 +0200 Subject: [FieldTrip] mixed between/within subject Anova In-Reply-To: <76DC57362137854CB9592B496182D5FA0506E253FF@wusmexmbx2.medpriv.wucon.wustl.edu> References: <76DC57362137854CB9592B496182D5FA0506E253FF@wusmexmbx2.medpriv.wucon.wustl.edu> Message-ID: <025001cc5d6a$b00dc230$10294690$@maris@donders.ru.nl> Dear Olga, > I'd like to use mixed between subjects (3 groups) and within subject (2 > factors each has 2 levels) Anova design to compare the ERPs using cluster > permutation statistics (ft_timelockananlysis). > But to my understanding it is impossible. Am I right? The problem is with the interaction effects. With permutation tests, you cannot test the interaction between two (or more) between-UO variables (UO=subject or trial) nor the interaction between two or more within-UO variables. However, with permutation tests, you can test the interaction between a between- and a within-UO variables. So. In your study, you would only be unable to test the interaction between the two within-subject variables. There have been more threads on the testing of interaction effects. You can find them by searching in the discussion list archive. Best, Eric Maris > > You comments are highly appreciated, > > Best Regards, > Olga. > > > Olga Sysoeva, > Research Associate, PhD > Washington University School of Medicine > Campus Box 8134 > 660 South Euclid Ave > Saint Louis, MO 63110-9909 > > The materials in this message are private and may contain Protected > Healthcare Information or other information of a sensitive nature. If you are > not the intended recipient, be advised that any unauthorized use, disclosure, > copying or the taking of any action in reliance on the contents of this > information is strictly prohibited. If you have received this email in error, > please immediately notify the sender via telephone or return mail. > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From michael.wibral at web.de Thu Aug 18 09:57:03 2011 From: michael.wibral at web.de (Michael Wibral) Date: Thu, 18 Aug 2011 09:57:03 +0200 (CEST) Subject: [FieldTrip] mixed between/within subject Anova Message-ID: <961614915.5812596.1313654223009.JavaMail.fmail@mwmweb069> Dear Olga, you might find this reference helpful with respect to permutation tests and ANOVA-like questions: PERMUTATION TESTS FOR MULTI-FACTORIAL ANALYSIS OF VARIANCE MARTI J. ANDERSON and CAJO J. F. TER BRAAK Journal of Statistical Computation and Simulation, 2003, Vol. 73(2), pp. 85–113 Michael (let me know if you have trouble getting it) -----Ursprüngliche Nachricht----- Von: "Eric Maris" Gesendet: Aug 18, 2011 7:50:08 AM An: "'Email discussion list for the FieldTrip project'" Betreff: Re: [FieldTrip] mixed between/within subject Anova >Dear Olga, > >> I'd like to use mixed between subjects (3 groups) and within subject (2 >> factors each has 2 levels) Anova design to compare the ERPs using cluster >> permutation statistics (ft_timelockananlysis). >> But to my understanding it is impossible. Am I right? > >The problem is with the interaction effects. With permutation tests, you >cannot test the interaction between two (or more) between-UO variables >(UO=subject or trial) nor the interaction between two or more within-UO >variables. However, with permutation tests, you can test the interaction >between a between- and a within-UO variables. So. In your study, you would >only be unable to test the interaction between the two within-subject >variables. > >There have been more threads on the testing of interaction effects. You can >find them by searching in the discussion list archive. > >Best, > >Eric Maris > > > > >> >> You comments are highly appreciated, >> >> Best Regards, >> Olga. >> >> >> Olga Sysoeva, >> Research Associate, PhD >> Washington University School of Medicine >> Campus Box 8134 >> 660 South Euclid Ave >> Saint Louis, MO 63110-9909 >> >> The materials in this message are private and may contain Protected >> Healthcare Information or other information of a sensitive nature. If you >are >> not the intended recipient, be advised that any unauthorized use, >disclosure, >> copying or the taking of any action in reliance on the contents of this >> information is strictly prohibited. If you have received this email in >error, >> please immediately notify the sender via telephone or return mail. >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > >_______________________________________________ >fieldtrip mailing list >fieldtrip at donders.ru.nl >http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From Margit.Schoenherr at uk-erlangen.de Thu Aug 18 11:15:25 2011 From: Margit.Schoenherr at uk-erlangen.de (=?iso-8859-1?Q?Sch=F6nherr=2C_Margit?=) Date: Thu, 18 Aug 2011 11:15:25 +0200 Subject: [FieldTrip] Forward solution using concentric spheres and BEM models In-Reply-To: References: Message-ID: <71D8F8A56F37A947912FC86D4D9F00F4633F43AE27@XMAIL1.medads.uk-erlangen.de> Hello, I probably cannot help you with all your questions, but currently I'm also dealing with MRI segmentation and volume conductors, so maybe my solutions can be useful for you. Concerning your point 5 (geometrical description of the brain, skull, and scalp), I have done it the following way: cfg = []; cfg.method = 'segmentation'; cfg.tissue = [1 2 3]; % scalp = 1; skull = 2; brain = 3; cfg.numvertices = [2000 1000 800]; cfg.sourceunits = 'mm'; cfg.mriunits = 'mm'; bnd = ft_prepare_mesh(cfg, mri_segment); I think you have to specify a vector 'cfg.numvertices', a number of vertices for each compartment. By the way, at the beginning of ft_prepare_mesh the cfg-field 'numcompartments' is removed from the cfg structure (at least in my fieldtrip version, which is not the latest). Another question that comes to me is, why you need the segmented MRI for a concentric spheres volume conductor? So, regarding point 6, you can build such a vol-structure for example like this: vol.unit = 'm'; vol.o = origin; vol.r = [R_brain R_skull R_skin]; vol.c = [1 1/80 1]; By this, you can also define the radius as you like. Best, Margit ________________________________________ Von: fieldtrip-bounces at donders.ru.nl [fieldtrip-bounces at donders.ru.nl] im Auftrag von Juan Pablo Neira [jpnv2006 at gmail.com] Gesendet: Mittwoch, 17. August 2011 14:58 An: fieldtrip at donders.ru.nl Betreff: [FieldTrip] Forward solution using concentric spheres and BEM models Hello, I am working with an individual MRI and EEG data to do the forward solution with two different volume's model (concentric spheres and BEM), so at the end i can compare the results. But i did not have the expected results of the concentric spheres model until now. This is my script: 1. Read individual MRI mri=ft_read_mri('data_patient_1\******.hdr'); %I got this "Warning: flipping 1st dimension (L-R) to obtain volume in neurological convention" %How can i do a homogenous transformation matrix, using the voxel dimensions %that are specified in hdr.dime.pixdim? 2. Realign to 'head coordinates' 3. MRI Reslicing 4. MRI segmentation: (brain, skull,scalp) 5. Create geometrical description of the brain, skull and scalp (3 compartments) cfg = []; cfg.interactive = 'no'; %segmentation method cfg.numvertices = 3000; cfg.sourceunits = 'mm'; cfg.downsample = 2; cfg.numcompartments = 3; cfg.tissue = {'brain' 'skull' 'scalp'}; cfg.smooth = 'yes'; bnd = ft_prepare_mesh(cfg, mri_segment); %The output just give me the geometrical description of the brain. How can i get the geometrical description of %the skull and scalp also to proceed to create a 3 spheres concentric sphere model? 6. If i have the 3 geometrical descriptions (brain, skull and scalp). Create a 3 concentric spheres model. cfg = []; cfg.headshape = [bnd1 bnd2 bnd3]; % brain- skull - scalp cfg.conductivity = [0.33 0.0042 0.331]; [vol, cfg] = prepare_concentricspheres(cfg); %I got the model but part of the brain is outside of its sphere and also the skull. How can I increase the radius so % the whole brain will be inside its sphere and also the skull? The sphere of the scalp fix good. I hope someone can help me solve these questions. Regards, Juan Pablo _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From litvak.vladimir at gmail.com Thu Aug 18 11:37:59 2011 From: litvak.vladimir at gmail.com (Vladimir Litvak) Date: Thu, 18 Aug 2011 10:37:59 +0100 Subject: [FieldTrip] bdf files persisting in memory? In-Reply-To: <028001cc5d3f$45e3edb0$d1abc910$%Budd@newcastle.edu.au> References: <9231708890694273.WA.cl304kent.ac.uk@jiscmail.ac.uk> <028001cc5d3f$45e3edb0$d1abc910$%Budd@newcastle.edu.au> Message-ID: Dear Bill, This might be a memory leek in the low-level mex file. It seems that the code for this file is available so it can be fixed. I'm CCing this to the Fieldtrip list and I'll also submit a bug report to Fieldtrip developers. Best, Vladimir On Thu, Aug 18, 2011 at 1:39 AM, Bill Budd wrote: > Dear Vladimir > > I've noticed when using spm_eeg_convert to convert bdf files that the bdf > file persists in memory after the spm_eeg_convert function or even the > script calling it has finished. Not sure if the issue is specific to my > scripting or the spm/fieldtrip functions. Wondered if you have any advice as > it uses a lot of memory while the rest of the script pipeline runs. The only > way I can seem to clear it is to run 'clear all mex' or exit matlab, but > really need some less sledge-hammer style solution to clearing it from RAM. > > Cheers > -Bill > > From drivolta81 at gmail.com Thu Aug 18 15:18:19 2011 From: drivolta81 at gmail.com (Davide Rivolta) Date: Thu, 18 Aug 2011 15:18:19 +0200 Subject: [FieldTrip] common filters - beamforming Message-ID: Dear Fieldtrippers, I am trying to learn beamforming using the tutorial. I am succesful in calculating, with my own data, the relative change between the stuimulus and the baseline condition (as shown in the tutorial). However, if I want to do statistic (ie., compare the baseline and stimulus condition) I would need a common filter (not in the tutorial). I found very useful the website where this step is explained (examples of matlab scripts section). I am however confused on how to perform stats at the source level. I get the 2 conditions separately and I wish to compare them and hopefully plot WHERE the difference is seen. Do you have any advice on how to do it (or where to look)? Here is my script and I am sure I am doing some silly mistake in the last (statistic part) since it gives the error: ??? Reference to non-existent field 'pow'. cfg = []; cfg.toilim = [-0.5 0]; dataPre = ft_redefinetrial(cfg, DataOut); cfg.toilim = [0.5 1]; dataPost = ft_redefinetrial(cfg, DataOut); data = ft_appenddata([], dataPre, dataPost); design = [ones(1, length(dataPre.trial)) ones(1, length(dataPost.trial))*2]; cfg = []; cfg.method = 'mtmfft'; cfg.output = 'powandcsd'; cfg.taper = 'dpss'; cfg.keeptrials = 'yes'; cfg.keeptapers = 'no'; cfg.foi = 57:2:63; cfg.tapsmofrq = ones(length(cfg.foi), 1).* 5; freq = ft_freqanalysis(cfg, data); cfg = []; cfg.frequency = 60; cfg.method = 'dics'; cfg.grid = grid; cfg.vol = vol; cfg.grad = DataOut.hdr.grad; cfg.keepfilter = 'yes'; source = ft_sourceanalysis(cfg, freq); cfg = []; cfg.frequency = 60; cfg.method = 'dics'; cfg.grid = grid; cfg.vol = vol; cfg.grad = DataOut.hdr.grad; cfg.grid.filter = source.avg.filter; %uses a precomputed filter (common filter) cfg.rawtrial = 'yes'; source = ft_sourceanalysis(cfg, freq); A = find(design==1); B = find(design==2); sourceA = source; sourceA.trial(B) = []; sourceA.cumtapcnt(B) = []; sourceA.df = length(A); cfg = []; cfg.keeptrials = 'yes'; sourceA = ft_sourcedescriptives([], sourceA); sourceB = source; sourceB.trial(A) = []; sourceB.cumtapcnt(A) = []; sourceB.df = length(B); cfg = []; cfg.keeptrials = 'yes'; sourceB = ft_sourcedescriptives([], sourceB); %STATS cfg = []; cfg.parameter = 'pow'; cfg.method = 'analytic'; cfg.statistic = 'actvsblT'; cfg.computestat = 'yes'; cfg.computecritval = 'yes'; cfg.computeprob = 'yes'; cfg.alpha = 0.05; cfg.tail = 0; stat = ft_sourcestatistics(cfg, sourceB, sourceA); Thanks for your help, Davide -------------- next part -------------- An HTML attachment was scrubbed... URL: From sysoevao at psychiatry.wustl.edu Thu Aug 18 18:02:11 2011 From: sysoevao at psychiatry.wustl.edu (Sysoeva, Olga Vladimirovna) Date: Thu, 18 Aug 2011 11:02:11 -0500 Subject: [FieldTrip] mixed between/within subject Anova In-Reply-To: <961614915.5812596.1313654223009.JavaMail.fmail@mwmweb069> References: <961614915.5812596.1313654223009.JavaMail.fmail@mwmweb069> Message-ID: <76DC57362137854CB9592B496182D5FA0506F6EAFB@wusmexmbx2.medpriv.wucon.wustl.edu> Thank you Michael, Thank you Eric Maris, So, as far as I understand there are some suggestion (Anderson et al, 2003) about how to apply the permutation test to my case (1 between and 2 within subjects factors), but it is not implemented in fieldtrip yet. As for examination the interaction for 1 between and 1 within subject factors - it is possible to do with fieldtrip scripts, isn't it? I've looked through the archive and one of the option seemed to be just to examine the between subject effect on the data obtained from subtraction of one level of within subject variable from the another one ( in case of 2 levels). And the results will represent the interaction effect. Am I right? Best Regards, Olga. -----Original Message----- From: fieldtrip-bounces at donders.ru.nl [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Michael Wibral Sent: Thursday, August 18, 2011 2:57 AM To: Email discussion list for the FieldTrip project Subject: Re: [FieldTrip] mixed between/within subject Anova Dear Olga, you might find this reference helpful with respect to permutation tests and ANOVA-like questions: PERMUTATION TESTS FOR MULTI-FACTORIAL ANALYSIS OF VARIANCE MARTI J. ANDERSON and CAJO J. F. TER BRAAK Journal of Statistical Computation and Simulation, 2003, Vol. 73(2), pp. 85–113 Michael (let me know if you have trouble getting it) -----Ursprüngliche Nachricht----- Von: "Eric Maris" Gesendet: Aug 18, 2011 7:50:08 AM An: "'Email discussion list for the FieldTrip project'" Betreff: Re: [FieldTrip] mixed between/within subject Anova >Dear Olga, > >> I'd like to use mixed between subjects (3 groups) and within subject >> (2 factors each has 2 levels) Anova design to compare the ERPs using >> cluster permutation statistics (ft_timelockananlysis). >> But to my understanding it is impossible. Am I right? > >The problem is with the interaction effects. With permutation tests, >you cannot test the interaction between two (or more) between-UO >variables (UO=subject or trial) nor the interaction between two or more >within-UO variables. However, with permutation tests, you can test the >interaction between a between- and a within-UO variables. So. In your >study, you would only be unable to test the interaction between the two >within-subject variables. > >There have been more threads on the testing of interaction effects. You >can find them by searching in the discussion list archive. > >Best, > >Eric Maris > > > > >> >> You comments are highly appreciated, >> >> Best Regards, >> Olga. >> >> >> Olga Sysoeva, >> Research Associate, PhD >> Washington University School of Medicine Campus Box 8134 >> 660 South Euclid Ave >> Saint Louis, MO 63110-9909 >> >> The materials in this message are private and may contain Protected >> Healthcare Information or other information of a sensitive nature. If >> you >are >> not the intended recipient, be advised that any unauthorized use, >disclosure, >> copying or the taking of any action in reliance on the contents of >> this information is strictly prohibited. If you have received this >> email in >error, >> please immediately notify the sender via telephone or return mail. >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > >_______________________________________________ >fieldtrip mailing list >fieldtrip at donders.ru.nl >http://mailman.science.ru.nl/mailman/listinfo/fieldtrip _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. From sust2001 at yahoo.ca Thu Aug 18 21:22:56 2011 From: sust2001 at yahoo.ca (Mark Taylor) Date: Thu, 18 Aug 2011 12:22:56 -0700 (PDT) Subject: [FieldTrip] Need your help to run *.ntt data file in the FieldTrip Message-ID: <1313695376.13817.YahooMailNeo@web46410.mail.sp1.yahoo.com> Hi, Could you please let me know whether I can  run the spike data file recorded with *.ntt format from Neuralynx in the FieldTrip environment? As this *.ntt format is not listed in the option of supported spike and LFP data formats, I would be very grateful if I can get your valuable suggestions to make the FieldTrip useful for my recorded data analysis. Your kind help and the earliest reply would be really appreciated. Thanking you, Mark. -------------- next part -------------- An HTML attachment was scrubbed... URL: From Ulrich.Pomper at charite.de Fri Aug 19 12:27:20 2011 From: Ulrich.Pomper at charite.de (Pomper, Ulrich) Date: Fri, 19 Aug 2011 12:27:20 +0200 Subject: [FieldTrip] =?windows-1252?q?1_PhD_and_1_Postdoctoral_Position_-_?= =?windows-1252?q?Charit=E9_Berlin?= Message-ID: The Department of Psychiatry and Psychotherapy, Charité – Universitätsmedizin Berlin (http://psy-ccm.charite.de/en/) invites applications for a Post-doctoral and a PhD student position. A Starting Grant of the European Research Council (ERC) will fund both positions for initially 2 years with the possibility of extension. The main objective of this ERC research program is to examine neural and neurochemical markers of multisensory integration and to test a new hypothesis that considers dynamic interplay of synchronized neural populations as a key to multisensory processes (Senkowski et al. 2008, Trends in Neurosciences). The studies within this program include healthy subjects and patients with schizophrenia, as a prototype of a mental disorder with deficits in multisensory integration. Multisensory processes will be examined in a series of experiments requiring both bottom-up and top-down processing (Talsma et al. 2010, Trends in Cognitive Sciences). Reaching beyond the state-of-the-art, this comprises a combination of human EEG/MEG data as a macroscopic measure of cortical processing; source modeling of synchronized EEG/MEG activity; and neurochemical markers using MR spectroscopy. Applicants should have a background in psychology, medicine, biology, physics or neuroscience. Experience in human EEG/MEG studies or biosignal analysis (e.g., Matlab) is desirable (i.e. required for the Post-doctoral position). An interest in neurophysiologic studies in clinical populations is expected, as well as basic German language skills for interacting with patients. Applicants are asked to submit their CV, a short motivation letter, 2 names of referees, and documentation of relevant qualification (e.g., copies of diplomas and/or transcripts of grades) until September 15, 2011, electronically to Dr. Daniel Senkowski (daniel.senkowski at charite.de; www.danielsenkowski.com). --------------------------------------------------------------- Daniel Senkowski, Ph.D Head of the research group Multisensory-Mind Department of Psychiatry and Psychotherapy Charité, University Medicine Berlin St. Hedwig Hospital, Große Hamburger Str. 5-11 10115 Berlin, Germany Phone: +49-30-2311-2738 Fax: +49-30-2311-2750 www.danielsenkowski.com From jpnv2006 at gmail.com Fri Aug 19 16:01:16 2011 From: jpnv2006 at gmail.com (Juan Pablo Neira) Date: Fri, 19 Aug 2011 16:01:16 +0200 Subject: [FieldTrip] Forward solution using concentric spheres and BEM models In-Reply-To: <71D8F8A56F37A947912FC86D4D9F00F4633F43AE27@XMAIL1.medads.uk-erlangen.de> References: <71D8F8A56F37A947912FC86D4D9F00F4633F43AE27@XMAIL1.medads.uk-erlangen.de> Message-ID: Thank you Margit for your help, I tried your sugestion to create the geometrical description of the brain, skull and scalp but I am still having the same output bnd = 1x1 struct (pnt and tri just of the brain). I should have bnd = 1x3 struct (pnt and tri of the brain, skull and scalp). This is the information in the command line in matlab not downsampling brain not downsampling scalp not downsampling skull using the segmentation approach using the segmented MRI triangulating the boundary of compartment 1 segmentation compartment 1 of 1 completed Can you tell me which version of fieldtrip (fieldtrip-yyyymmdd) are you using so i could compare the code of the function ft_prepare_mesh. Regards, Juan Pablo 2011/8/18 Schönherr, Margit : > Hello, > > I probably cannot help you with all your questions, but currently I'm also dealing with MRI segmentation and volume conductors, so maybe my solutions can be useful for you. > Concerning your point 5 (geometrical description of the brain, skull, and scalp), I have done it the following way: > > cfg             = []; > cfg.method      = 'segmentation'; > cfg.tissue      = [1 2 3]; % scalp = 1; skull = 2; brain = 3; > cfg.numvertices = [2000 1000 800]; > cfg.sourceunits = 'mm'; > cfg.mriunits    = 'mm'; > bnd = ft_prepare_mesh(cfg, mri_segment); > > I think you have to specify a vector 'cfg.numvertices', a number of vertices for each compartment. > By the way, at the beginning of ft_prepare_mesh the cfg-field 'numcompartments' is removed from the cfg structure (at least in my fieldtrip version, which is not the latest). > > Another question that comes to me is, why you need the segmented MRI for a concentric spheres volume conductor? So, regarding point 6, you can build such a vol-structure for example like this: > > vol.unit = 'm'; > vol.o = origin; > vol.r = [R_brain R_skull R_skin]; > vol.c = [1 1/80 1]; > > By this, you can also define the radius as you like. > > Best, > Margit > > > > > ________________________________________ > Von: fieldtrip-bounces at donders.ru.nl [fieldtrip-bounces at donders.ru.nl] im Auftrag von Juan Pablo Neira [jpnv2006 at gmail.com] > Gesendet: Mittwoch, 17. August 2011 14:58 > An: fieldtrip at donders.ru.nl > Betreff: [FieldTrip] Forward solution using concentric spheres and BEM models > > Hello, > > I am working with an individual MRI and EEG data to do the forward > solution with two different volume's model (concentric spheres and > BEM), > so at the end i can compare the results.  But i did not have the > expected results of the concentric spheres model until now. > > This is my script: > > 1.  Read individual MRI > mri=ft_read_mri('data_patient_1\******.hdr'); > > %I got this "Warning: flipping 1st dimension (L-R) to obtain volume in > neurological convention" > > %How can i do a homogenous transformation matrix, using the voxel dimensions > %that are specified in hdr.dime.pixdim? > > 2. Realign to 'head coordinates' > > 3. MRI Reslicing > > 4. MRI segmentation: (brain, skull,scalp) > > 5. Create geometrical description of the brain, skull and scalp (3 compartments) > > cfg                 = []; > cfg.interactive     = 'no';     %segmentation method > cfg.numvertices     = 3000; > cfg.sourceunits     = 'mm'; > cfg.downsample      = 2; > cfg.numcompartments = 3; > cfg.tissue          = {'brain' 'skull' 'scalp'}; > cfg.smooth          = 'yes'; > bnd                 = ft_prepare_mesh(cfg, mri_segment); > > %The output just give me the geometrical description of the brain. > How can i get the geometrical description of > %the skull and scalp also to proceed to create a 3 spheres concentric > sphere model? > > 6.  If i have the 3 geometrical descriptions (brain, skull and scalp). >  Create a 3 concentric spheres model. > > cfg                 = []; > cfg.headshape       = [bnd1 bnd2 bnd3];   % brain- skull - scalp > cfg.conductivity    = [0.33 0.0042 0.331]; > [vol, cfg]          = prepare_concentricspheres(cfg); > > %I got the model but part of the brain is outside of its sphere and > also the skull.  How can I increase the radius so > % the whole brain will be inside its sphere and also the skull?  The > sphere of the scalp fix good. > > I hope someone can help me solve these questions. > > Regards, > > Juan Pablo > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > From Margit.Schoenherr at uk-erlangen.de Fri Aug 19 17:23:34 2011 From: Margit.Schoenherr at uk-erlangen.de (=?iso-8859-1?Q?Sch=F6nherr=2C_Margit?=) Date: Fri, 19 Aug 2011 17:23:34 +0200 Subject: [FieldTrip] Forward solution using concentric spheres and BEM models In-Reply-To: References: <71D8F8A56F37A947912FC86D4D9F00F4633F43AE27@XMAIL1.medads.uk-erlangen.de>, Message-ID: <71D8F8A56F37A947912FC86D4D9F00F4633F43AE2A@XMAIL1.medads.uk-erlangen.de> Hello Juan Pablo, I am using fieldtrip-20110603. Now that you say, that your problem persisted, I remember having changed the code of prepare_mesh_segmentation. You find the function in the private directory. In line 25, it sets cfg.tissue = 1 if mri has fields 'brain' or 'scalp'. I think this explains why the segmentation is done for compartment 1 only. So I have simply commented this line, so that cfg.tissue is left unchanged, and the segmentation is done for all compartments. I know that this is bad style, but I wanted to get it working. And since I was relatively new to fieldtrip, I hesitated to ask the discussion list... I hope this helps. And I'm open for better solutions. Best, Margit ________________________________________ Von: fieldtrip-bounces at donders.ru.nl [fieldtrip-bounces at donders.ru.nl] im Auftrag von Juan Pablo Neira [jpnv2006 at gmail.com] Gesendet: Freitag, 19. August 2011 16:01 An: Email discussion list for the FieldTrip project Betreff: Re: [FieldTrip] Forward solution using concentric spheres and BEM models Thank you Margit for your help, I tried your sugestion to create the geometrical description of the brain, skull and scalp but I am still having the same output bnd = 1x1 struct (pnt and tri just of the brain). I should have bnd = 1x3 struct (pnt and tri of the brain, skull and scalp). This is the information in the command line in matlab not downsampling brain not downsampling scalp not downsampling skull using the segmentation approach using the segmented MRI triangulating the boundary of compartment 1 segmentation compartment 1 of 1 completed Can you tell me which version of fieldtrip (fieldtrip-yyyymmdd) are you using so i could compare the code of the function ft_prepare_mesh. Regards, Juan Pablo 2011/8/18 Schönherr, Margit : > Hello, > > I probably cannot help you with all your questions, but currently I'm also dealing with MRI segmentation and volume conductors, so maybe my solutions can be useful for you. > Concerning your point 5 (geometrical description of the brain, skull, and scalp), I have done it the following way: > > cfg = []; > cfg.method = 'segmentation'; > cfg.tissue = [1 2 3]; % scalp = 1; skull = 2; brain = 3; > cfg.numvertices = [2000 1000 800]; > cfg.sourceunits = 'mm'; > cfg.mriunits = 'mm'; > bnd = ft_prepare_mesh(cfg, mri_segment); > > I think you have to specify a vector 'cfg.numvertices', a number of vertices for each compartment. > By the way, at the beginning of ft_prepare_mesh the cfg-field 'numcompartments' is removed from the cfg structure (at least in my fieldtrip version, which is not the latest). > > Another question that comes to me is, why you need the segmented MRI for a concentric spheres volume conductor? So, regarding point 6, you can build such a vol-structure for example like this: > > vol.unit = 'm'; > vol.o = origin; > vol.r = [R_brain R_skull R_skin]; > vol.c = [1 1/80 1]; > > By this, you can also define the radius as you like. > > Best, > Margit > > > > > ________________________________________ > Von: fieldtrip-bounces at donders.ru.nl [fieldtrip-bounces at donders.ru.nl] im Auftrag von Juan Pablo Neira [jpnv2006 at gmail.com] > Gesendet: Mittwoch, 17. August 2011 14:58 > An: fieldtrip at donders.ru.nl > Betreff: [FieldTrip] Forward solution using concentric spheres and BEM models > > Hello, > > I am working with an individual MRI and EEG data to do the forward > solution with two different volume's model (concentric spheres and > BEM), > so at the end i can compare the results. But i did not have the > expected results of the concentric spheres model until now. > > This is my script: > > 1. Read individual MRI > mri=ft_read_mri('data_patient_1\******.hdr'); > > %I got this "Warning: flipping 1st dimension (L-R) to obtain volume in > neurological convention" > > %How can i do a homogenous transformation matrix, using the voxel dimensions > %that are specified in hdr.dime.pixdim? > > 2. Realign to 'head coordinates' > > 3. MRI Reslicing > > 4. MRI segmentation: (brain, skull,scalp) > > 5. Create geometrical description of the brain, skull and scalp (3 compartments) > > cfg = []; > cfg.interactive = 'no'; %segmentation method > cfg.numvertices = 3000; > cfg.sourceunits = 'mm'; > cfg.downsample = 2; > cfg.numcompartments = 3; > cfg.tissue = {'brain' 'skull' 'scalp'}; > cfg.smooth = 'yes'; > bnd = ft_prepare_mesh(cfg, mri_segment); > > %The output just give me the geometrical description of the brain. > How can i get the geometrical description of > %the skull and scalp also to proceed to create a 3 spheres concentric > sphere model? > > 6. If i have the 3 geometrical descriptions (brain, skull and scalp). > Create a 3 concentric spheres model. > > cfg = []; > cfg.headshape = [bnd1 bnd2 bnd3]; % brain- skull - scalp > cfg.conductivity = [0.33 0.0042 0.331]; > [vol, cfg] = prepare_concentricspheres(cfg); > > %I got the model but part of the brain is outside of its sphere and > also the skull. How can I increase the radius so > % the whole brain will be inside its sphere and also the skull? The > sphere of the scalp fix good. > > I hope someone can help me solve these questions. > > Regards, > > Juan Pablo > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From P.Praamstra at neuro.umcn.nl Fri Aug 19 21:04:29 2011 From: P.Praamstra at neuro.umcn.nl (P.Praamstra at neuro.umcn.nl) Date: Fri, 19 Aug 2011 21:04:29 +0200 Subject: [FieldTrip] PhD and post-doctoral position at Donders Centre for Cognitive Neuroimaging Message-ID: Please find attached two adverts for positions at the Donders Centre for Cognitive Neuroimaging, Radboud University Nijmegen, The Netherlands. For both positions we are looking for candidates with MEG or EEG experience. Post-doctoral researcher on project "Movement selection in the parietofrontal cortex: The affordance competition hypothesis". PhD student for project "Feeling the beat: The neurophysiology of cueing in Parkinson's disease". Peter Praamstra, MD, PhD Department of Neurology Radboud University Nijmegen Medical Centre PO Box 9101, 6500 HB Nijmegen The Netherlands Tel: 00-31-24-3668254 Fax: 00-31-24-3541122 Email: p.praamstra at neuro.umcn.nl Het UMC St Radboud staat geregistreerd bij de Kamer van Koophandel in het handelsregister onder nummer 41055629. The Radboud University Nijmegen Medical Centre is listed in the Commercial Register of the Chamber of Commerce under file number 41055629. -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: Advert_Post-doc.doc Type: application/msword Size: 33280 bytes Desc: Advert_Post-doc.doc URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: Advert_AiO.doc Type: application/msword Size: 33280 bytes Desc: Advert_AiO.doc URL: From lmoranr at gmail.com Tue Aug 23 12:02:33 2011 From: lmoranr at gmail.com (=?ISO-8859-1?Q?Luis_Mor=E1n?=) Date: Tue, 23 Aug 2011 12:02:33 +0200 Subject: [FieldTrip] Doubt about de units Message-ID: Dear fieldtrippers, I'm doing a simulation with somato data from MEG. I know that measured fields in data are in* femtoteslas*. Well, the procedure is: - Load data - After setting the configuration structure "cfg", I make a dipolesimulation, and then timelockanalysis, i.e.: *raw1 = dipolesimulation(cfg) avg1=timelockanalysis(cfg, raw1)* What I wonder to know is which are the resultant units of the field * avg1.avg* In the same way, then I perform a reconstruction using beamforming: *sourceAvg1 = sourceanalysis(cfg_beamforming,avg1);* And also, I would like to know the units in which the field * sourceAvg1.avg.pow* is measured Thanks in advance Regards -- Luis Morán Rodríguez -------------- next part -------------- An HTML attachment was scrubbed... URL: From Antony.Passaro at uth.tmc.edu Tue Aug 23 20:50:47 2011 From: Antony.Passaro at uth.tmc.edu (Passaro, Antony D) Date: Tue, 23 Aug 2011 13:50:47 -0500 Subject: [FieldTrip] Source Timecourse In-Reply-To: References: Message-ID: Hi Fieldtrippers, I was looking for a way to extract a time-course from each source estimated by the lcmv beamformer. I searched through the mailing list archives and came across a few emails which talked about the possibility but I was unable to find any examples explaining exactly how this process is done. I am interested in extracting the time-course over the average response rather than the individual trials, if possible. Any help would be greatly appreciated. Thanks, -Tony -------------- next part -------------- An HTML attachment was scrubbed... URL: From Don.Rojas at ucdenver.edu Thu Aug 25 19:43:03 2011 From: Don.Rojas at ucdenver.edu (Rojas, Don) Date: Thu, 25 Aug 2011 11:43:03 -0600 Subject: [FieldTrip] Source Timecourse In-Reply-To: References: Message-ID: <51AF637A-B1EB-4421-84B3-6B0D0C4BBA24@ucdenver.edu> Tony, I don't use the LCMV beamformer, so I can't tell you what the structure of the output is, but what you need to perform a source space projection is the orientation and location information for the sources. In the DICS beamformer output, which I'm familiar with, this information would be in the source structure as follows (assuming your source output structure is called "source"): source.pos % the position info as an n x 3 matrix, where n is the number of sources, in the same units as your volume conductor model source.avg.ori % the orientation as an 1 x n cell array, where n is the number of locations and each cell is a 3 x 1 array of x y z orientation and n refers to the % same source in each array So, in that case, you might find the necessary information for the peak source, in my case, as follows: [~,ind]=max(source.avg.pow(:)); % index of the source with the maximum power in the dics output, let's say ind = 5000 ori=source.avg.ori(5000); % orientation info loc=source.pos(5000,:); % location information Then, if you get this info, you can search the archives for my posts, one of which has a detailed explanation for getting the source waveform when you have a known source location, source orientation and a volume conductor. If you want to get all the waveforms, you can of course iterate the process over the entire voxel volume, but since the leadfields are highly correlated in adjacent source locations, you will see that many of the waveforms will look essentially the same for sources within a few centimeters. Best, Don ----------------------- Don Rojas, Ph.D. Associate Professor of Psychiatry U. of Colorado Denver Anschutz Medical Campus Director, UCD Magnetoencephalography Lab 13001 E. 17th Pl F546 Aurora, CO 80045 USA 303-724-4994 On Aug 23, 2011, at 12:50 PM, Passaro, Antony D wrote: Hi Fieldtrippers, I was looking for a way to extract a time-course from each source estimated by the lcmv beamformer. I searched through the mailing list archives and came across a few emails which talked about the possibility but I was unable to find any examples explaining exactly how this process is done. I am interested in extracting the time-course over the average response rather than the individual trials, if possible. Any help would be greatly appreciated. Thanks, -Tony _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip -------------- next part -------------- An HTML attachment was scrubbed... URL: From ekanal at cmu.edu Fri Aug 26 18:19:45 2011 From: ekanal at cmu.edu (Kanal Eliezer) Date: Fri, 26 Aug 2011 12:19:45 -0400 Subject: [FieldTrip] assessing significance in using ft_timelockanalysis results Message-ID: <8FC0FE65-8E0D-4D7C-9A62-CE1610845C92@cmu.edu> Hello folks - The event related statistics tutorial (http://fieldtrip.fcdonders.nl/tutorial/eventrelatedstatistics) talks about assessing significance parametrically by running t-tests on pooled timelockanalysis data. My question is, does the fact that the averages were created from N trials make a difference? If I'm condition A has twelve averages and condition B has another twelve, and each average contains 70 trials, is there a way to "inform" the statistical test that the power in this dataset is greater than 24? Is this only possible if I run the t-test comparing each set of 840 (70*12) trials? I'm also curious whether this is possible with non-parametric analyses, as well. Thanks - Elli Kanal -------------------- Eliezer Kanal, Ph.D. Postdoctoral Fellow Center for the Neural Basis of Cognition Carnegie Mellon University 4400 Fifth Ave, Suite 110A Pittsburgh PA 15213 P: 412-268-4115 F: 412-268-5060 From e.maris at donders.ru.nl Fri Aug 26 21:00:18 2011 From: e.maris at donders.ru.nl (Eric Maris) Date: Fri, 26 Aug 2011 21:00:18 +0200 Subject: [FieldTrip] assessing significance in using ft_timelockanalysis results In-Reply-To: <8FC0FE65-8E0D-4D7C-9A62-CE1610845C92@cmu.edu> References: <8FC0FE65-8E0D-4D7C-9A62-CE1610845C92@cmu.edu> Message-ID: <020701cc6422$65b10f30$31132d90$@maris@donders.ru.nl> Hi Kanal, > The event related statistics tutorial > (http://fieldtrip.fcdonders.nl/tutorial/eventrelatedstatistics) talks about > assessing significance parametrically by running t-tests on pooled > timelockanalysis data. My question is, does the fact that the averages were > created from N trials make a difference? If I'm condition A has twelve > averages and condition B has another twelve, and each average contains 70 > trials, is there a way to "inform" the statistical test that the power in this > dataset is greater than 24? Is this only possible if I run the t-test comparing > each set of 840 (70*12) trials? > > I'm also curious whether this is possible with non-parametric analyses, as > well. Thanks - In an analysis over subjects (called random-effects analysis in the fMRI literature), "informing" the statistical test about the number of trials per condition only makes sense if this number is different for the two conditions. I propose that you have a look the fMRI papers that deal with the issue of fixed-versus-random effect analyses. The conceptual issues involved are the same in fMRI and electrophysiology. Best, Eric Maris > > Elli Kanal > > > -------------------- > Eliezer Kanal, Ph.D. > Postdoctoral Fellow > Center for the Neural Basis of Cognition > Carnegie Mellon University > 4400 Fifth Ave, Suite 110A > Pittsburgh PA 15213 > P: 412-268-4115 > F: 412-268-5060 > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From tim.curran at Colorado.EDU Sat Aug 27 21:31:25 2011 From: tim.curran at Colorado.EDU (Tim Curran) Date: Sat, 27 Aug 2011 13:31:25 -0600 Subject: [FieldTrip] Fwd: Postdoctoral Position Job Announcement for Bellugi at Salk Institute References: <34FBC147-2620-42D5-ACE8-0D5FD88AC5BD@eng.ucsd.edu> Message-ID: <1686575C-9B5C-4ACB-B752-EF360CF1B8B0@colorado.edu> > > > > -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: Job Ad Final Version_06_14_11.pdf Type: application/pdf Size: 728679 bytes Desc: Job Ad Final Version_06_14_11.pdf URL: From r.oostenveld at donders.ru.nl Mon Aug 29 13:24:21 2011 From: r.oostenveld at donders.ru.nl (Robert Oostenveld) Date: Mon, 29 Aug 2011 13:24:21 +0200 Subject: [FieldTrip] Source Timecourse In-Reply-To: <51AF637A-B1EB-4421-84B3-6B0D0C4BBA24@ucdenver.edu> References: <51AF637A-B1EB-4421-84B3-6B0D0C4BBA24@ucdenver.edu> Message-ID: Dear Tony The lcmv beamformer (by default) estimates the time series of the source, given that the source has an unknown orientation. The result is a 3xNtime matrix as the source activation, which is in source.avg.mom (or in source.trial(i).mom). The three rows of the moment are the strength of the source in the x-, y- and z-direction. Using ft_sourcedescriptives with the option cfg.projectmom=yes you can project the source to its strongest orientation, i.e. the direction that explains most of the source variance. That is equivalent to taking the largest eigenvector of the source timeseries (which is a phrasing that is often used in papers, also on combining fMRI bold timeseries over multiple voxels). You can also do it manually, like this mom3d = randn(3,100); % example source dipole moment [u, s, v] = svd(mom, 'econ'); mom1d = v(1,:); Alternative to taking the largest eigenvector (which has a sign-ambiguity), you can do something like mom3d = randn(3,100); % example source dipole moment [u, s, v] = svd(mom, 'econ'); mom1d = abs(v(1,:)); to take the absolute value, or mom3d = randn(3,100); % example source dipole moment mom1d = sqrt(sum(mom3d.^2,1)); to take the strength over all three directions for each timepoint. best Robert On 25 Aug 2011, at 19:43, Rojas, Don wrote: > Tony, > > I don't use the LCMV beamformer, so I can't tell you what the structure of the output is, but what you need to perform a source space projection is the orientation and location information for the sources. In the DICS beamformer output, which I'm familiar with, this information would be in the source structure as follows (assuming your source output structure is called "source"): > > source.pos % the position info as an n x 3 matrix, where n is the number of sources, in the same units as your volume conductor model > source.avg.ori % the orientation as an 1 x n cell array, where n is the number of locations and each cell is a 3 x 1 array of x y z orientation and n refers to the % same source in each array > > So, in that case, you might find the necessary information for the peak source, in my case, as follows: > > [~,ind]=max(source.avg.pow(:)); % index of the source with the maximum power in the dics output, let's say ind = 5000 > ori=source.avg.ori(5000); % orientation info > loc=source.pos(5000,:); % location information > > Then, if you get this info, you can search the archives for my posts, one of which has a detailed explanation for getting the source waveform when you have a known source location, source orientation and a volume conductor. If you want to get all the waveforms, you can of course iterate the process over the entire voxel volume, but since the leadfields are highly correlated in adjacent source locations, you will see that many of the waveforms will look essentially the same for sources within a few centimeters. > > Best, > > Don > ----------------------- > Don Rojas, Ph.D. > Associate Professor of Psychiatry > U. of Colorado Denver Anschutz Medical Campus > Director, UCD Magnetoencephalography Lab > 13001 E. 17th Pl F546 > Aurora, CO 80045 USA > 303-724-4994 > > On Aug 23, 2011, at 12:50 PM, Passaro, Antony D wrote: > >> Hi Fieldtrippers, >> >> I was looking for a way to extract a time-course from each source estimated by the lcmv beamformer. I searched through the mailing list archives and came across a few emails which talked about the possibility but I was unable to find any examples explaining exactly how this process is done. I am interested in extracting the time-course over the average response rather than the individual trials, if possible. Any help would be greatly appreciated. >> >> Thanks, >> -Tony >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From Antony.Passaro at uth.tmc.edu Mon Aug 29 18:33:55 2011 From: Antony.Passaro at uth.tmc.edu (Passaro, Antony D) Date: Mon, 29 Aug 2011 11:33:55 -0500 Subject: [FieldTrip] Source Timecourse In-Reply-To: References: <51AF637A-B1EB-4421-84B3-6B0D0C4BBA24@ucdenver.edu> Message-ID: Hi Robert and Don, Thank you both for your feedback, I really appreciate it. Using the ft_sourcedescriptives function (and projectmom), I was able to extract the time-course for each source. I am still curious about the output though as it appears to represent the entire duration (including the baseline) of the epoch rather than the specified time-window defined during timelockanalysis (just prior to sourceanalysis). That being the case, does this time-course represent the .pow estimated by the lcmv beamformer or is the nai? Or is it simply a source representation of the gradiometers over the entire epoch? Another issue I came across concerns the source interpolation. It seems once I interpolate the sources using a template mri, only the pow and nai fields are interpolated....is there a way to also interpolate the mom field as well or is it a matter of backtracking to figure out which original source corresponds to each interpolated source? Thank you again for your help, -Tony -----Original Message----- From: fieldtrip-bounces at donders.ru.nl [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Robert Oostenveld Sent: Monday, August 29, 2011 6:24 AM To: Email discussion list for the FieldTrip project Subject: Re: [FieldTrip] Source Timecourse Dear Tony The lcmv beamformer (by default) estimates the time series of the source, given that the source has an unknown orientation. The result is a 3xNtime matrix as the source activation, which is in source.avg.mom (or in source.trial(i).mom). The three rows of the moment are the strength of the source in the x-, y- and z-direction. Using ft_sourcedescriptives with the option cfg.projectmom=yes you can project the source to its strongest orientation, i.e. the direction that explains most of the source variance. That is equivalent to taking the largest eigenvector of the source timeseries (which is a phrasing that is often used in papers, also on combining fMRI bold timeseries over multiple voxels). You can also do it manually, like this mom3d = randn(3,100); % example source dipole moment [u, s, v] = svd(mom, 'econ'); mom1d = v(1,:); Alternative to taking the largest eigenvector (which has a sign-ambiguity), you can do something like mom3d = randn(3,100); % example source dipole moment [u, s, v] = svd(mom, 'econ'); mom1d = abs(v(1,:)); to take the absolute value, or mom3d = randn(3,100); % example source dipole moment mom1d = sqrt(sum(mom3d.^2,1)); to take the strength over all three directions for each timepoint. best Robert On 25 Aug 2011, at 19:43, Rojas, Don wrote: > Tony, > > I don't use the LCMV beamformer, so I can't tell you what the structure of the output is, but what you need to perform a source space projection is the orientation and location information for the sources. In the DICS beamformer output, which I'm familiar with, this information would be in the source structure as follows (assuming your source output structure is called "source"): > > source.pos % the position info as an n x 3 matrix, where n is the number of sources, in the same units as your volume conductor model > source.avg.ori % the orientation as an 1 x n cell array, where n is the number of locations and each cell is a 3 x 1 array of x y z orientation and n refers to the % same source in each array > > So, in that case, you might find the necessary information for the peak source, in my case, as follows: > > [~,ind]=max(source.avg.pow(:)); % index of the source with the maximum power in the dics output, let's say ind = 5000 > ori=source.avg.ori(5000); % orientation info > loc=source.pos(5000,:); % location information > > Then, if you get this info, you can search the archives for my posts, one of which has a detailed explanation for getting the source waveform when you have a known source location, source orientation and a volume conductor. If you want to get all the waveforms, you can of course iterate the process over the entire voxel volume, but since the leadfields are highly correlated in adjacent source locations, you will see that many of the waveforms will look essentially the same for sources within a few centimeters. > > Best, > > Don > ----------------------- > Don Rojas, Ph.D. > Associate Professor of Psychiatry > U. of Colorado Denver Anschutz Medical Campus Director, UCD > Magnetoencephalography Lab > 13001 E. 17th Pl F546 > Aurora, CO 80045 USA > 303-724-4994 > > On Aug 23, 2011, at 12:50 PM, Passaro, Antony D wrote: > >> Hi Fieldtrippers, >> >> I was looking for a way to extract a time-course from each source estimated by the lcmv beamformer. I searched through the mailing list archives and came across a few emails which talked about the possibility but I was unable to find any examples explaining exactly how this process is done. I am interested in extracting the time-course over the average response rather than the individual trials, if possible. Any help would be greatly appreciated. >> >> Thanks, >> -Tony >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From irina.simanova at mpi.nl Tue Aug 30 08:03:29 2011 From: irina.simanova at mpi.nl (Irina Simanova) Date: Tue, 30 Aug 2011 08:03:29 +0200 Subject: [FieldTrip] ICA on MEG data Message-ID: <729C68C8-BB67-44F7-B439-A5B9C90BBB9D@mpi.nl> Dear FieldTripers, I would like to identify ocular artifacts in an MEG dataset using the ICA. The data consist of 180 trials of variable duration ( the signal from stimulus onset to subject's response). The trials' duration vary from 400 ms to 3000 ms. The data is sampled at 1.2 kHz, and has 275 channels. I am new to ICA, and I need a piece of advice. What would be a reasonable number of output independent components? Should I do the PCA as a pre-processing stage? How do i find an optimal number of principal components to extract? I would appreciate your help, Kind regards, Irina Simanova PhD student Neurobiology of Language Group Max-Planck Institute for Psycholinguistics Wundtlaan 1 6525 XD Nijmegen The Netherlands e-mail: irina.simanova at mpi.nl phone: +31 24 3521541 From r.oostenveld at donders.ru.nl Tue Aug 30 09:08:19 2011 From: r.oostenveld at donders.ru.nl (Robert Oostenveld) Date: Tue, 30 Aug 2011 09:08:19 +0200 Subject: [FieldTrip] Source Timecourse In-Reply-To: References: <51AF637A-B1EB-4421-84B3-6B0D0C4BBA24@ucdenver.edu> Message-ID: <9370AD38-AA89-4038-957B-28525CAA0A37@donders.ru.nl> Hi Tony On 29 Aug 2011, at 18:33, Passaro, Antony D wrote: > Thank you both for your feedback, I really appreciate it. Using the ft_sourcedescriptives function (and projectmom), I was able to extract the time-course for each source. I am still curious about the output though as it appears to represent the entire duration (including the baseline) of the epoch rather than the specified time-window defined during timelockanalysis (just prior to sourceanalysis). Default behaviour is to construct the covariance and average for the whole timewindow, indeed including the baseline period. Depending on the experimental data, the LCMV filter can be optimized for a specific timewindow by computing the covariance only for that timewindow. Subsequently the filter can be applied to the whole timewindow (including baseline) or only to the timewindow of interest. > That being the case, does this time-course represent the .pow estimated by the lcmv beamformer or is the nai? source.avg.pow is the power estimated based on the data covariance window, not on the longer time window that you might pass through the filter. If you computed the covariance for a smaller timewindow (using ft_timelockanalysis), then only that is used for the power estimate. The same applies to the nai. > Or is it simply a source representation of the gradiometers over the entire epoch? Another issue I came across concerns the source interpolation. It seems once I interpolate the sources using a template mri, only the pow and nai fields are interpolated....is there a way to also interpolate the mom field as well or is it a matter of backtracking to figure out which original source corresponds to each interpolated source? The mom timecourse is indeed not interpolated, as that used to exceed the memory on many computers: the interpolated mom would consist of a full 3D volume at each timepoint of your ERF. We are in the process of making some improvements to the source data structure, i.e. to its representation. The planning is mostly done, but implementation wise I don't know how far the code already is. I think you could work with the following to interpolate it source.avg.momint = nan(length(source.pos,1), length(source.time)); for i=1:length(source.inside) indx = source.inside(i); source.avg.momint(indx,:) = source.avg.mom{indx}; % the grid locations outside the brain keep their NaN end and then specify cfg.parameter = 'momint' in ft_sourceinterpolate. best Robert From stan.vanpelt at fcdonders.ru.nl Tue Aug 30 09:14:31 2011 From: stan.vanpelt at fcdonders.ru.nl (Stan van Pelt) Date: Tue, 30 Aug 2011 09:14:31 +0200 (CEST) Subject: [FieldTrip] ICA on MEG data In-Reply-To: <729C68C8-BB67-44F7-B439-A5B9C90BBB9D@mpi.nl> References: <729C68C8-BB67-44F7-B439-A5B9C90BBB9D@mpi.nl> Message-ID: <0db501cc66e4$76b60a00$64221e00$@vanpelt@fcdonders.ru.nl> Dear Irina, There's a page on the Fieldtrip Wiki about using ICA to remove EOG-artifacts, see http://fieldtrip.fcdonders.nl/example/use_independent_component_analysis_i ca_to_remove_eog_artifacts. The common approach is to first define your trials, and then run ICA on them. You will get as many output components as you have channels, i.e. 275 in your case. The EOG artifacts usually pop up in 1 or 2 components within the first 20 largest components. Best regards, Stan Stan van Pelt, PhD Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen Kapittelweg 29, 6525 EN Nijmegen, Netherlands E-mail: stan.vanpelt at donders.ru.nl Website: www.ru.nl/donders/ Phone: (+31) (0)24 36 68495 Fax: (+31) (0)24 36 10989 -----Original Message----- From: fieldtrip-bounces at donders.ru.nl [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Irina Simanova Sent: Tuesday, August 30, 2011 8:03 AM To: Email discussion list for the FieldTrip project Subject: [FieldTrip] ICA on MEG data Dear FieldTripers, I would like to identify ocular artifacts in an MEG dataset using the ICA. The data consist of 180 trials of variable duration ( the signal from stimulus onset to subject's response). The trials' duration vary from 400 ms to 3000 ms. The data is sampled at 1.2 kHz, and has 275 channels. I am new to ICA, and I need a piece of advice. What would be a reasonable number of output independent components? Should I do the PCA as a pre-processing stage? How do i find an optimal number of principal components to extract? I would appreciate your help, Kind regards, Irina Simanova PhD student Neurobiology of Language Group Max-Planck Institute for Psycholinguistics Wundtlaan 1 6525 XD Nijmegen The Netherlands e-mail: irina.simanova at mpi.nl phone: +31 24 3521541 _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From irina.simanova at mpi.nl Tue Aug 30 09:56:12 2011 From: irina.simanova at mpi.nl (Irina Simanova) Date: Tue, 30 Aug 2011 09:56:12 +0200 Subject: [FieldTrip] ICA on MEG data In-Reply-To: <0db501cc66e4$76b60a00$64221e00$@vanpelt@fcdonders.ru.nl> References: <729C68C8-BB67-44F7-B439-A5B9C90BBB9D@mpi.nl> <0db501cc66e4$76b60a00$64221e00$@vanpelt@fcdonders.ru.nl> Message-ID: <79A92C58-4FA2-4ADD-B6D6-975A4C4B9F08@mpi.nl> Dear Stan, Thank you for your reply, As far as I understand from the tutorials about ICA by Scott Makeig http://sccn.ucsd.edu/~scott/tutorial/icafaq.html , the number of time points should be considerably larger than the number of scalp sensors. How critical is this point? Besides, the sensors are very close to each other, and the registered activity in adjacent sensors is highly correlated, so the components would be similar, right? Isn't it reasonable to reduce the number of output components then? Thank you in advance, Irina On Aug 30, 2011, at 9:14 AM 8/30/11, Stan van Pelt wrote: > Dear Irina, > > There's a page on the Fieldtrip Wiki about using ICA to remove > EOG-artifacts, see > http://fieldtrip.fcdonders.nl/example/use_independent_component_analysis_i > ca_to_remove_eog_artifacts. > > The common approach is to first define your trials, and then run ICA > on > them. > > You will get as many output components as you have channels, i.e. > 275 in > your case. The EOG artifacts usually pop up in 1 or 2 components > within > the first 20 largest components. > > Best regards, > Stan > > Stan van Pelt, PhD > Donders Institute for Brain, Cognition and Behaviour, Radboud > University > Nijmegen > Kapittelweg 29, 6525 EN Nijmegen, Netherlands > E-mail: stan.vanpelt at donders.ru.nl > Website: www.ru.nl/donders/ > Phone: (+31) (0)24 36 68495 > Fax: (+31) (0)24 36 10989 > > -----Original Message----- > From: fieldtrip-bounces at donders.ru.nl > [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Irina Simanova > Sent: Tuesday, August 30, 2011 8:03 AM > To: Email discussion list for the FieldTrip project > Subject: [FieldTrip] ICA on MEG data > > Dear FieldTripers, > > I would like to identify ocular artifacts in an MEG dataset using the > ICA. > > The data consist of 180 trials of variable duration ( the signal from > stimulus onset to subject's response). The trials' duration vary from > 400 ms to 3000 ms. The data is sampled at 1.2 kHz, and has 275 > channels. I am new to ICA, and I need a piece of advice. What would be > a reasonable number of output independent components? > Should I do the PCA as a pre-processing stage? How do i find an > optimal number of principal components to extract? > > I would appreciate your help, > > Kind regards, > > Irina Simanova > PhD student > Neurobiology of Language Group > Max-Planck Institute for Psycholinguistics > Wundtlaan 1 > 6525 XD Nijmegen > The Netherlands > e-mail: irina.simanova at mpi.nl > phone: +31 24 3521541 > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From stan.vanpelt at fcdonders.ru.nl Tue Aug 30 10:15:07 2011 From: stan.vanpelt at fcdonders.ru.nl (Stan van Pelt) Date: Tue, 30 Aug 2011 10:15:07 +0200 (CEST) Subject: [FieldTrip] ICA on MEG data In-Reply-To: <79A92C58-4FA2-4ADD-B6D6-975A4C4B9F08@mpi.nl> References: <729C68C8-BB67-44F7-B439-A5B9C90BBB9D@mpi.nl> <0db501cc66e4$76b60a00$64221e00$@vanpelt@fcdonders.ru.nl> <79A92C58-4FA2-4ADD-B6D6-975A4C4B9F08@mpi.nl> Message-ID: <0e0601cc66ec$ee2048f0$ca60dad0$@vanpelt@fcdonders.ru.nl> Dear Irina, The number of time points should indeed be much larger for a good ICA estimate. However, that seems to be the case with your data, so I think there's no problem there. As far as I know, with ICA in fieldtrip you always get the same amount of output components as input channels. Activity in adjacent MEG channels will be somewhat correlated indeed, but not nearly as much as with EEG. Moreover, the origin of your artifacts (EOG, maybe also ECG) will create a distinct topographic distribution, and the components will pop out clearly from your ft_componentanalysis, because it explains large amounts of variance in the data. Just give it a try, I'd say. You should get similar results as in the Wiki example. Best, Stan -----Original Message----- From: fieldtrip-bounces at donders.ru.nl [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Irina Simanova Sent: Tuesday, August 30, 2011 9:56 AM To: Email discussion list for the FieldTrip project Subject: Re: [FieldTrip] ICA on MEG data Dear Stan, Thank you for your reply, As far as I understand from the tutorials about ICA by Scott Makeig http://sccn.ucsd.edu/~scott/tutorial/icafaq.html , the number of time points should be considerably larger than the number of scalp sensors. How critical is this point? Besides, the sensors are very close to each other, and the registered activity in adjacent sensors is highly correlated, so the components would be similar, right? Isn't it reasonable to reduce the number of output components then? Thank you in advance, Irina On Aug 30, 2011, at 9:14 AM 8/30/11, Stan van Pelt wrote: > Dear Irina, > > There's a page on the Fieldtrip Wiki about using ICA to remove > EOG-artifacts, see > http://fieldtrip.fcdonders.nl/example/use_independent_component_analysis_i > ca_to_remove_eog_artifacts. > > The common approach is to first define your trials, and then run ICA > on > them. > > You will get as many output components as you have channels, i.e. > 275 in > your case. The EOG artifacts usually pop up in 1 or 2 components > within > the first 20 largest components. > > Best regards, > Stan > > Stan van Pelt, PhD > Donders Institute for Brain, Cognition and Behaviour, Radboud > University > Nijmegen > Kapittelweg 29, 6525 EN Nijmegen, Netherlands > E-mail: stan.vanpelt at donders.ru.nl > Website: www.ru.nl/donders/ > Phone: (+31) (0)24 36 68495 > Fax: (+31) (0)24 36 10989 > > -----Original Message----- > From: fieldtrip-bounces at donders.ru.nl > [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Irina Simanova > Sent: Tuesday, August 30, 2011 8:03 AM > To: Email discussion list for the FieldTrip project > Subject: [FieldTrip] ICA on MEG data > > Dear FieldTripers, > > I would like to identify ocular artifacts in an MEG dataset using the > ICA. > > The data consist of 180 trials of variable duration ( the signal from > stimulus onset to subject's response). The trials' duration vary from > 400 ms to 3000 ms. The data is sampled at 1.2 kHz, and has 275 > channels. I am new to ICA, and I need a piece of advice. What would be > a reasonable number of output independent components? > Should I do the PCA as a pre-processing stage? How do i find an > optimal number of principal components to extract? > > I would appreciate your help, > > Kind regards, > > Irina Simanova > PhD student > Neurobiology of Language Group > Max-Planck Institute for Psycholinguistics > Wundtlaan 1 > 6525 XD Nijmegen > The Netherlands > e-mail: irina.simanova at mpi.nl > phone: +31 24 3521541 > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip _______________________________________________ fieldtrip mailing list fieldtrip at donders.ru.nl http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From k.muesch at uke.uni-hamburg.de Tue Aug 30 10:44:42 2011 From: k.muesch at uke.uni-hamburg.de (=?iso-8859-1?Q?Kathrin_M=FCsch?=) Date: Tue, 30 Aug 2011 10:44:42 +0200 Subject: [FieldTrip] ICA on MEG data In-Reply-To: <0e0601cc66ec$ee2048f0$ca60dad0$@vanpelt@fcdonders.ru.nl> References: <729C68C8-BB67-44F7-B439-A5B9C90BBB9D@mpi.nl> <0db501cc66e4$76b60a00$64221e00$@vanpelt@fcdonders.ru.nl> <79A92C58-4FA2-4ADD-B6D6-975A4C4B9F08@mpi.nl> <0e0601cc66ec$ee2048f0$ca60dad0$@vanpelt@fcdonders.ru.nl> Message-ID: Dear Irina, I reduced the number of components to 64 by running a PCA beforehand (cfg.runica.pca = 64) on my MEG data. Depending on how much the subjects moved their eyes, blinks and saccades popped up in earlier or later components. I had feasible results with this approach both for EOG and ECG. Good luck, Kathrin _____________________________________ Kathrin Müsch Dept. of Neurophysiology and Pathophysiology University Medical Center Hamburg-Eppendorf Martinistr. 52 20246 Hamburg Germany Phone: +49-40-7410-54680 Fax: +49-40-7410-57752 E-Mail: k.muesch at uke.uni-hamburg.de _____________________________________ Am 30.08.2011 um 10:15 schrieb Stan van Pelt: > Dear Irina, > > The number of time points should indeed be much larger for a good ICA > estimate. However, that seems to be the case with your data, so I think > there's no problem there. > > As far as I know, with ICA in fieldtrip you always get the same amount of > output components as input channels. Activity in adjacent MEG channels > will be somewhat correlated indeed, but not nearly as much as with EEG. > Moreover, the origin of your artifacts (EOG, maybe also ECG) will create a > distinct topographic distribution, and the components will pop out clearly > from your ft_componentanalysis, because it explains large amounts of > variance in the data. > > Just give it a try, I'd say. You should get similar results as in the Wiki > example. > > Best, > Stan > > > -----Original Message----- > From: fieldtrip-bounces at donders.ru.nl > [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Irina Simanova > Sent: Tuesday, August 30, 2011 9:56 AM > To: Email discussion list for the FieldTrip project > Subject: Re: [FieldTrip] ICA on MEG data > > Dear Stan, > > Thank you for your reply, > > As far as I understand from the tutorials about ICA by Scott Makeig > http://sccn.ucsd.edu/~scott/tutorial/icafaq.html > , the number of time points should be considerably larger than the > number of scalp sensors. How critical is this point? > > Besides, the sensors are very close to each other, and the registered > activity in adjacent sensors is highly correlated, so the components > would be similar, right? Isn't it reasonable to reduce the number of > output components then? > > Thank you in advance, > Irina > > > > On Aug 30, 2011, at 9:14 AM 8/30/11, Stan van Pelt wrote: > >> Dear Irina, >> >> There's a page on the Fieldtrip Wiki about using ICA to remove >> EOG-artifacts, see >> > http://fieldtrip.fcdonders.nl/example/use_independent_component_analysis_i >> ca_to_remove_eog_artifacts. >> >> The common approach is to first define your trials, and then run ICA >> on >> them. >> >> You will get as many output components as you have channels, i.e. >> 275 in >> your case. The EOG artifacts usually pop up in 1 or 2 components >> within >> the first 20 largest components. >> >> Best regards, >> Stan >> >> Stan van Pelt, PhD >> Donders Institute for Brain, Cognition and Behaviour, Radboud >> University >> Nijmegen >> Kapittelweg 29, 6525 EN Nijmegen, Netherlands >> E-mail: stan.vanpelt at donders.ru.nl >> Website: www.ru.nl/donders/ >> Phone: (+31) (0)24 36 68495 >> Fax: (+31) (0)24 36 10989 >> >> -----Original Message----- >> From: fieldtrip-bounces at donders.ru.nl >> [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Irina Simanova >> Sent: Tuesday, August 30, 2011 8:03 AM >> To: Email discussion list for the FieldTrip project >> Subject: [FieldTrip] ICA on MEG data >> >> Dear FieldTripers, >> >> I would like to identify ocular artifacts in an MEG dataset using the >> ICA. >> >> The data consist of 180 trials of variable duration ( the signal from >> stimulus onset to subject's response). The trials' duration vary from >> 400 ms to 3000 ms. The data is sampled at 1.2 kHz, and has 275 >> channels. I am new to ICA, and I need a piece of advice. What would be >> a reasonable number of output independent components? >> Should I do the PCA as a pre-processing stage? How do i find an >> optimal number of principal components to extract? >> >> I would appreciate your help, >> >> Kind regards, >> >> Irina Simanova >> PhD student >> Neurobiology of Language Group >> Max-Planck Institute for Psycholinguistics >> Wundtlaan 1 >> 6525 XD Nijmegen >> The Netherlands >> e-mail: irina.simanova at mpi.nl >> phone: +31 24 3521541 >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From irina.simanova at mpi.nl Tue Aug 30 14:22:31 2011 From: irina.simanova at mpi.nl (Irina Simanova) Date: Tue, 30 Aug 2011 14:22:31 +0200 Subject: [FieldTrip] Call for Papers :: NIPS 2011 Workshop on Interpretable Decoding of Higher Cognitive States from Neural Data Message-ID: <2E7CF2CC-C7AF-4937-935E-D0337E8445C8@mpi.nl> Interpretable Decoding of Higher Cognitive States from Neural Data NIPS 2011 Workshop, Dec 16 or 17, 2011, Granada, Spain (Please feel free to distribute the CFP to all the interested persons and groups.) Overview Over recent years, machine learning methods have become a crucial analytical tool in cognitive neuroscience (see reviews by Formisano et al., 2008; Pereira et al., 2009). Decoding techniques have dramatically increased the sensitivity of experiments, and so also the subtlety of cognitive questions that can be asked. At the same time the mental phenomena being studied are moving beyond lower-level perceptual and motor processes which are directly grounded in external measurable realities. Decoding higher cognition and interpreting the learned behaviour of the classifiers used pose unique challenges, as these psychological states are complex, fast-changing and often ill-defined. Contemporary machine learning methods deal well with the small numbers of cases, and high numbers of co-linear dimensions typical of neural data, and are generally optimized to maximize classification performance, rather than to enable meaningful interpretation of the features they learn from. And indeed recent work has succeeded to decode psychological phenomena including visual object recognition (e.g. Kriegeskorte et al., 2008; Connolly et al., 2011), perceptual interpretation of sounds (Staeren et al., 2009), lexical semantics (Mitchell et al., 2008; Siminova et al., 2010; Devereux et al., 2010; Murphy et al., 2011), decision making during game playing (Xiang et al., 2009) and the process of mental arithmetic (Anderson et al., 2008). But for the cognitive scientists who use these methods, the primary question is often not "how much" but rather "how" and "why" the patterns of neural activity identified by a machine learning algorithm encode particular cognitive processes. The aim of this workshop is therefore to 1) discuss the achievements and problems of the decoding of high-level cognitive states, and 2) explore the use of machine learning methodologies and other computational models that enable such cognitive interpretation of neural recordings of different modalities. Advances in this field require close collaboration between machine learning experts, neuroscientists and cognitive scientists. Thus, this workshop is highly interdisciplinary and will aim to attract submissions also from outside the existing NIPS community. By stimulating discussions among experts in the different fields, the workshop seeks to generate novel insights and new directions for research. Topics of interest The field requires techniques that are capable of taking advantage of spatially distributed patterns in the brain, that are separated in space but coordinated in their activity. Methods should also be sensitive to the fine-grained temporal patterns of multiple processes - which may proceed in a serial fashion, overlapping or in parallel with each-other, or in multiple passes with bidirectional information flows. Different recording modalities have distinctive advantages: fMRI provides very fine millimetre-level localisation in the brain but poor temporal resolution, while EEG and MEG have millisecond temporal resolution at the cost of spatial resolution. Ideally machine learning methods would be able to meaningfully combine complementary information from these different neuroimaging techniques (see e.g. De Martino et al., 2010). Moreover, as the processes underlying higher cognition are so complex, methods should be able to disentangle even tightly linked and confounded subprocesses. Finally, general use algorithms that could induce latent dimensions from neural data, and so reveal the "hidden" psychological states, would be a dramatic advance on current hypothesis-driven analytical paradigms. Originality of approach is encouraged and submissions on any related methodological approach are welcomed, such as: - Interpreting spatial and temporal location of selected features and their weights - Discovering "hidden" or "latent" cognitive representations - Disentangling confounded processes and representations - Comparing or combining data from recording modalities (e.g. fMRI, EEG, structural MRI, DTI, MEG, NIRS, EcOG, single cell recordings) - Fuzzy and partial classifications - Unaligned or incommensurate feature spaces and data representation As noted above, the complexity of higher cognition poses challenges. To take language comprehension as an example, speech is received at 3-4 words; acoustic, semantic and syntactic processing can occur in parallel; and the form of underlying representations (sentence structures, conceptual descriptions) remains controversial. We welcome submissions dealing with any high-level cognitive functions that exhibit similar complexity, for instance: - Knowledge representation and concepts - Language and communication - Understanding visual and auditory experience - Memory and learning - Reasoning and problem solving - Decision making and executive control Submissions Authors are invited to submit full papers on original, unpublished work in the topic area of this workshop via the NIPS 2011 submission site at https://cmt.research.microsoft.com/NIPS2011/Default.aspx. Submissions should be formatted using the NIPS 2011 stylefiles, with blind review and not exceeding 8 pages plus an extra page for references. Author and submission information can be found at http://nips.cc/PaperInformation/AuthorSubmissionInstructions . The stylefiles are available at http://nips.cc/PaperInformation/StyleFiles . Each submission will be reviewed at least by two members of the programme committee. Accepted papers will be published in the workshop proceedings. Dual submissions to the main NIPS 2011 conference and this workshop are allowed; if you submit to the main session, indicate this when you submit to the workshop. If your paper is accepted for the main session, you should withdraw your paper from the workshop upon notification by the main session. Important Dates - Aug 30, 2011: Call for papers - Sep 23, 2011: Deadline for submission of workshop papers - Oct 15, 2011: Notification of acceptance - Oct 31, 2011: Camera-ready papers due - Dec 16 or 17, 2011: Workshop date Links - NIPS 2011 website: http://nips.cc/Conferences/2011/ - Workshop website: https://sites.google.com/site/decodehighcogstate - Call for Papers: https://sites.google.com/site/decodehighcogstate/cfp/ Kind regards, The Workshop Organizers, Kai-min Kevin Chang, Anna Korhonen, Irina Simanova, Brian Murphy -------------- next part -------------- An HTML attachment was scrubbed... URL: From anna.lambrechts at gmail.com Tue Aug 30 17:46:52 2011 From: anna.lambrechts at gmail.com (Anna Lambrechts) Date: Tue, 30 Aug 2011 17:46:52 +0200 Subject: [FieldTrip] Issues with neighbours in ft_timelockstatistics Message-ID: Dear fieldtrip community, Until about two weeks ago I was using a batch to process group comparison statistics on MEG data using ft_timelockstatistics. A week later, the same batch was not working anymore, even on the same data. The error I get is the following: *??? Improper index matrix reference. Error in ==> clusterstat>makechanneighbstructmat at 520 [seld] = match_str(cfg.channel, cfg.neighbours(chan).label); Error in ==> clusterstat at 60 channeighbstructmat = makechanneighbstructmat(cfg); Error in ==> statistics_montecarlo at 320 [stat, cfg] = clusterstat(cfg, statrand, statobs,'issource',issource); Error in ==> statistics_wrapper at 290 [stat, cfg] = statmethod(cfg, dat, design, 'issource',issource); Error in ==> ft_timelockstatistics at 124 [stat, cfg] = statistics_wrapper(cfg, varargin{:}); Error in ==> myfunction_stats at 59 gpstat = ft_timelockstatistics(cfg, data1, data2); *It seems the error concerns the 'neighbours.mat' file reading, even though I am not sure about it. When I delete the two last lines in this file (to match the size of matrix wanted) the error disappears, but the processing is incomplete and false (or at least it does not match the results I had before). Has anyone encounter the same issue? Did you do any update recently that might explain this problem, and do you have any idea how to remedy to it? Best wishes, Anna. -------------- next part -------------- An HTML attachment was scrubbed... URL: From ekanal at cmu.edu Tue Aug 30 17:58:34 2011 From: ekanal at cmu.edu (Kanal Eliezer) Date: Tue, 30 Aug 2011 11:58:34 -0400 Subject: [FieldTrip] assessing significance in using ft_timelockanalysis results In-Reply-To: <2109_1314385277_p7QJ1Gjs003376_020701cc6422$65b10f30$31132d90$@maris@donders.ru.nl> References: <8FC0FE65-8E0D-4D7C-9A62-CE1610845C92@cmu.edu> <2109_1314385277_p7QJ1Gjs003376_020701cc6422$65b10f30$31132d90$@maris@donders.ru.nl> Message-ID: <8CC6BA25-9A03-4630-B940-486BE2B9EE92@cmu.edu> Hello Eric - Thanks for the response. It looks like the fixed vs random analysis is exactly what I'm referring to. From what I understood, it looks like the difference really only shows up in the variance of the resultant distribution; with a random, the variance also takes into account the betwee-subjects variance. Is there a way to specify whether I want to do a fixed or random effects analysis in FieldTrip when I'm running ft_timelockgrandaverage or ft_freqgrandaverage? Thanks! Elli p.s. - In case anyone else is trying to figure this out, chapter 12 of Friston's book "Statistical Parametric Mapping" does an excellent job explaining the difference between fixed and random analyses, as well as how to implement it algorithmically. On Aug 26, 2011, at 3:00 PM, Eric Maris wrote: > Hi Kanal, > >> The event related statistics tutorial >> (http://fieldtrip.fcdonders.nl/tutorial/eventrelatedstatistics) talks > about >> assessing significance parametrically by running t-tests on pooled >> timelockanalysis data. My question is, does the fact that the averages > were >> created from N trials make a difference? If I'm condition A has twelve >> averages and condition B has another twelve, and each average contains 70 >> trials, is there a way to "inform" the statistical test that the power in > this >> dataset is greater than 24? Is this only possible if I run the t-test > comparing >> each set of 840 (70*12) trials? >> >> I'm also curious whether this is possible with non-parametric analyses, as >> well. Thanks - > > In an analysis over subjects (called random-effects analysis in the fMRI > literature), "informing" the statistical test about the number of trials per > condition only makes sense if this number is different for the two > conditions. I propose that you have a look the fMRI papers that deal with > the issue of fixed-versus-random effect analyses. The conceptual issues > involved are the same in fMRI and electrophysiology. > > > Best, > > Eric Maris > > > > >> >> Elli Kanal >> >> >> -------------------- >> Eliezer Kanal, Ph.D. >> Postdoctoral Fellow >> Center for the Neural Basis of Cognition >> Carnegie Mellon University >> 4400 Fifth Ave, Suite 110A >> Pittsburgh PA 15213 >> P: 412-268-4115 >> F: 412-268-5060 >> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From e.maris at donders.ru.nl Tue Aug 30 21:13:42 2011 From: e.maris at donders.ru.nl (Eric Maris) Date: Tue, 30 Aug 2011 21:13:42 +0200 Subject: [FieldTrip] assessing significance in using ft_timelockanalysis results Message-ID: <00d201cc6748$f18e8fe0$d4abafa0$@maris@donders.ru.nl> Hi Elli, > Thanks for the response. It looks like the fixed vs random analysis is exactly > what I'm referring to. From what I understood, it looks like the difference > really only shows up in the variance of the resultant distribution; with a > random, the variance also takes into account the betwee-subjects variance. > Is there a way to specify whether I want to do a fixed or random effects > analysis in FieldTrip when I'm running ft_timelockgrandaverage or > ft_freqgrandaverage? Thanks! I don't get this. Ft_timelockgrandaverage and ft_freqgrandaverage will only be called when a random effects analysis is performed. Best, Eric > > Elli > > p.s. - In case anyone else is trying to figure this out, chapter 12 of Friston's > book "Statistical Parametric Mapping" does an excellent job explaining the > difference between fixed and random analyses, as well as how to implement > it algorithmically. > > > > On Aug 26, 2011, at 3:00 PM, Eric Maris wrote: > > > Hi Kanal, > > > >> The event related statistics tutorial > >> (http://fieldtrip.fcdonders.nl/tutorial/eventrelatedstatistics) talks > > about > >> assessing significance parametrically by running t-tests on pooled > >> timelockanalysis data. My question is, does the fact that the averages > > were > >> created from N trials make a difference? If I'm condition A has twelve > >> averages and condition B has another twelve, and each average contains > 70 > >> trials, is there a way to "inform" the statistical test that the power in > > this > >> dataset is greater than 24? Is this only possible if I run the t-test > > comparing > >> each set of 840 (70*12) trials? > >> > >> I'm also curious whether this is possible with non-parametric analyses, as > >> well. Thanks - > > > > In an analysis over subjects (called random-effects analysis in the fMRI > > literature), "informing" the statistical test about the number of trials per > > condition only makes sense if this number is different for the two > > conditions. I propose that you have a look the fMRI papers that deal with > > the issue of fixed-versus-random effect analyses. The conceptual issues > > involved are the same in fMRI and electrophysiology. > > > > > > Best, > > > > Eric Maris > > > > > > > > > >> > >> Elli Kanal > >> > >> > >> -------------------- > >> Eliezer Kanal, Ph.D. > >> Postdoctoral Fellow > >> Center for the Neural Basis of Cognition > >> Carnegie Mellon University > >> 4400 Fifth Ave, Suite 110A > >> Pittsburgh PA 15213 > >> P: 412-268-4115 > >> F: 412-268-5060 > >> > >> > >> _______________________________________________ > >> fieldtrip mailing list > >> fieldtrip at donders.ru.nl > >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > _______________________________________________ > > fieldtrip mailing list > > fieldtrip at donders.ru.nl > > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From ekanal at cmu.edu Tue Aug 30 21:46:24 2011 From: ekanal at cmu.edu (Kanal Eliezer) Date: Tue, 30 Aug 2011 15:46:24 -0400 Subject: [FieldTrip] assessing significance in using ft_timelockanalysis results In-Reply-To: <29602_1314731679_p7UJEcWB023970_00d201cc6748$f18e8fe0$d4abafa0$@maris@donders.ru.nl> References: <29602_1314731679_p7UJEcWB023970_00d201cc6748$f18e8fe0$d4abafa0$@maris@donders.ru.nl> Message-ID: <4D99A829-9BB1-4395-9997-45568224DAEC@cmu.edu> Maybe I had been using it incorrectly, then. In my study, I have numerous datafiles for each subject (each block gets split into it's own .fif file, due to file size), and I use those functions to combine the preprocessed and timelocked datafiles. Should I be doing this differently? Elli On Aug 30, 2011, at 3:13 PM, Eric Maris wrote: > Hi Elli, > > > >> Thanks for the response. It looks like the fixed vs random analysis is > exactly >> what I'm referring to. From what I understood, it looks like the > difference >> really only shows up in the variance of the resultant distribution; with a >> random, the variance also takes into account the betwee-subjects variance. >> Is there a way to specify whether I want to do a fixed or random effects >> analysis in FieldTrip when I'm running ft_timelockgrandaverage or >> ft_freqgrandaverage? Thanks! > > > I don't get this. Ft_timelockgrandaverage and ft_freqgrandaverage will only > be called when a random effects analysis is performed. > > > Best, > > Eric > > >> >> Elli >> >> p.s. - In case anyone else is trying to figure this out, chapter 12 of > Friston's >> book "Statistical Parametric Mapping" does an excellent job explaining the >> difference between fixed and random analyses, as well as how to implement >> it algorithmically. >> >> >> >> On Aug 26, 2011, at 3:00 PM, Eric Maris wrote: >> >>> Hi Kanal, >>> >>>> The event related statistics tutorial >>>> (http://fieldtrip.fcdonders.nl/tutorial/eventrelatedstatistics) talks >>> about >>>> assessing significance parametrically by running t-tests on pooled >>>> timelockanalysis data. My question is, does the fact that the averages >>> were >>>> created from N trials make a difference? If I'm condition A has twelve >>>> averages and condition B has another twelve, and each average contains >> 70 >>>> trials, is there a way to "inform" the statistical test that the power > in >>> this >>>> dataset is greater than 24? Is this only possible if I run the t-test >>> comparing >>>> each set of 840 (70*12) trials? >>>> >>>> I'm also curious whether this is possible with non-parametric analyses, > as >>>> well. Thanks - >>> >>> In an analysis over subjects (called random-effects analysis in the fMRI >>> literature), "informing" the statistical test about the number of trials > per >>> condition only makes sense if this number is different for the two >>> conditions. I propose that you have a look the fMRI papers that deal > with >>> the issue of fixed-versus-random effect analyses. The conceptual issues >>> involved are the same in fMRI and electrophysiology. >>> >>> >>> Best, >>> >>> Eric Maris >>> >>> >>> >>> >>>> >>>> Elli Kanal >>>> >>>> >>>> -------------------- >>>> Eliezer Kanal, Ph.D. >>>> Postdoctoral Fellow >>>> Center for the Neural Basis of Cognition >>>> Carnegie Mellon University >>>> 4400 Fifth Ave, Suite 110A >>>> Pittsburgh PA 15213 >>>> P: 412-268-4115 >>>> F: 412-268-5060 >>>> >>>> >>>> _______________________________________________ >>>> fieldtrip mailing list >>>> fieldtrip at donders.ru.nl >>>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >>> >>> _______________________________________________ >>> fieldtrip mailing list >>> fieldtrip at donders.ru.nl >>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip >> >> >> _______________________________________________ >> fieldtrip mailing list >> fieldtrip at donders.ru.nl >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip From semrich at wisc.edu Tue Aug 30 21:49:03 2011 From: semrich at wisc.edu (Stephen Emrich) Date: Tue, 30 Aug 2011 14:49:03 -0500 Subject: [FieldTrip] depsamplesregrT help Message-ID: Dear fieldtrip users and developers, I am attempting to use the depsamplesregrT statistic, but have encountered some problems. The documentation also appears to be incomplete. I am attempting to regress the outputs of ft_freqanalysis for 3 conditions against some values (the IVs). From what I can tell, depsamplesregrT should do this. Some issues I'm running into: cvar: cvar is not documented in this function, but i'm assuming from some searches that the cvar should be another row of values that actually correspond to the regressors (the IVs). If I attempt to run it without specifying cvar, I get an error. unitselved: even if I include the cvar in the design matrix, there is an error that 'unitselved' is undefined. I can't find any documentation on what this variable is or what the values should be. Any help would be appreciated. S- From jonas at obleser.de Wed Aug 31 12:10:40 2011 From: jonas at obleser.de (Jonas Obleser) Date: Wed, 31 Aug 2011 12:10:40 +0200 Subject: [FieldTrip] depsamplesregrT help Message-ID: <67EC1721-BF53-450B-A1DB-67F16F4E2CC7@obleser.de> Hi Stephen, from my understanding (greatly facilitated by Eric Maris a year ago or so), depsamplesregrT does only test for linear increases or decreases in your data, depending on the sequence in which you input your data conditions. For example, freqstatistics(cfg, cond1{:}, cond2{:}, cond3{:}) will result in positive clusters for data increasing (quasi-)linearly from cond1 to cond3, and negative clusters for data behaving vice versa. For regression/correlation with an independent (e.g., behavioural) regressor, we use a statfun Nathan Weisz (also active on the list here) once wrote and kindly shared with us. Hope this helps somewhat. Best wishes, J. Dr. Jonas Obleser Auditory Cognition Group Max Planck Institute for Human Cognitive and Brain Sciences Leipzig, Germany (p) +49 (0)341 9940 114 (e) obleser at cbs.mpg.de From jm.horschig at donders.ru.nl Wed Aug 31 13:02:52 2011 From: jm.horschig at donders.ru.nl (=?ISO-8859-1?Q?=22J=F6rn_M=2E_Horschig=22?=) Date: Wed, 31 Aug 2011 13:02:52 +0200 Subject: [FieldTrip] Issues with neighbours in ft_timelockstatistics In-Reply-To: References: Message-ID: <4E5E14DC.8050000@donders.ru.nl> Hi Anna, As you might have read, we recently updated how neighbours are computed and stored, and that you have to specify your neighbours manually now. Could you let me know how you defined the neighbours and how many channels your data has? Maybe there is a mismatch between the neighbour struct and your data in number of sensors, but that's just a wild guess. Anyhow, the result might look different depending on how you define your neighbours. Anyway, if you provide more information I can look into this and see how to fix it. Best, Jörn On 8/30/2011 5:46 PM, Anna Lambrechts wrote: > Dear fieldtrip community, > > Until about two weeks ago I was using a batch to process group > comparison statistics on MEG data using ft_timelockstatistics. A week > later, the same batch was not working anymore, even on the same data. > > The error I get is the following: > > *??? Improper index matrix reference. > > Error in ==> clusterstat>makechanneighbstructmat at 520 > [seld] = match_str(cfg.channel, cfg.neighbours(chan).label); > > Error in ==> clusterstat at 60 > channeighbstructmat = makechanneighbstructmat(cfg); > > Error in ==> statistics_montecarlo at 320 > [stat, cfg] = clusterstat(cfg, statrand, statobs,'issource',issource); > > Error in ==> statistics_wrapper at 290 > [stat, cfg] = statmethod(cfg, dat, design, 'issource',issource); > > Error in ==> ft_timelockstatistics at 124 > [stat, cfg] = statistics_wrapper(cfg, varargin{:}); > > Error in ==> myfunction_stats at 59 > gpstat = ft_timelockstatistics(cfg, data1, data2); > > *It seems the error concerns the 'neighbours.mat' file reading, even > though I am not sure about it. When I delete the two last lines in > this file (to match the size of matrix wanted) the error disappears, > but the processing is incomplete and false (or at least it does not > match the results I had before). > > Has anyone encounter the same issue? > Did you do any update recently that might explain this problem, and do > you have any idea how to remedy to it? > > Best wishes, > Anna. > > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip -- Jörn M. Horschig PhD Student Donders Institute for Brain, Cognition and Behaviour Centre for Cognitive Neuroimaging Radboud University Nijmegen Neuronal Oscillations Group P.O. Box 9101 NL-6500 HB Nijmegen The Netherlands Contact: E-Mail: jm.horschig at donders.ru.nl Tel: +31-(0)24-36-68493 Web: http://www.ru.nl/donders Visiting address: Trigon, room 2.30 Kapittelweg 29 NL-6525 EN Nijmegen The Netherlands -------------- next part -------------- An HTML attachment was scrubbed... URL: From anna.lambrechts at gmail.com Wed Aug 31 13:40:33 2011 From: anna.lambrechts at gmail.com (Anna Lambrechts) Date: Wed, 31 Aug 2011 13:40:33 +0200 Subject: [FieldTrip] fieldtrip Digest, Vol 9, Issue 30 In-Reply-To: References: Message-ID: Hello, I have written yesterday about troubles while running ft_timelockstatistics. As it happens now, this seems to be a compatibility issue between matlab 2011 and the current fieldtrip version (which you can solve temporarily by using older versions of both). Best wishes, Anna. 2011/8/30 > Send fieldtrip mailing list submissions to > fieldtrip at donders.ru.nl > > To subscribe or unsubscribe via the World Wide Web, visit > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > or, via email, send a message with subject or body 'help' to > fieldtrip-request at donders.ru.nl > > You can reach the person managing the list at > fieldtrip-owner at donders.ru.nl > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of fieldtrip digest..." > > > Today's Topics: > > 1. Call for Papers :: NIPS 2011 Workshop on Interpretable > Decoding of Higher Cognitive States from Neural Data (Irina Simanova) > 2. Issues with neighbours in ft_timelockstatistics (Anna Lambrechts) > 3. Re: assessing significance in using ft_timelockanalysis > results (Kanal Eliezer) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Tue, 30 Aug 2011 14:22:31 +0200 > From: Irina Simanova > To: Email discussion list for the FieldTrip project > > Subject: [FieldTrip] Call for Papers :: NIPS 2011 Workshop on > Interpretable Decoding of Higher Cognitive States from Neural Data > Message-ID: <2E7CF2CC-C7AF-4937-935E-D0337E8445C8 at mpi.nl> > Content-Type: text/plain; charset="us-ascii"; Format="flowed"; > DelSp="yes" > > Interpretable Decoding of Higher Cognitive States from Neural Data > > NIPS 2011 Workshop, Dec 16 or 17, 2011, Granada, Spain > > (Please feel free to distribute the CFP to all the interested persons > and groups.) > > Overview > > Over recent years, machine learning methods have become a crucial > analytical tool in cognitive neuroscience (see reviews by Formisano et > al., 2008; Pereira et al., 2009). Decoding techniques have > dramatically increased the sensitivity of experiments, and so also the > subtlety of cognitive questions that can be asked. At the same time > the mental phenomena being studied are moving beyond lower-level > perceptual and motor processes which are directly grounded in external > measurable realities. > > Decoding higher cognition and interpreting the learned behaviour of > the classifiers used pose unique challenges, as these psychological > states are complex, fast-changing and often ill-defined. Contemporary > machine learning methods deal well with the small numbers of cases, > and high numbers of co-linear dimensions typical of neural data, and > are generally optimized to maximize classification performance, rather > than to enable meaningful interpretation of the features they learn > from. And indeed recent work has succeeded to decode psychological > phenomena including visual object recognition (e.g. Kriegeskorte et > al., 2008; Connolly et al., 2011), perceptual interpretation of sounds > (Staeren et al., 2009), lexical semantics (Mitchell et al., 2008; > Siminova et al., 2010; Devereux et al., 2010; Murphy et al., 2011), > decision making during game playing (Xiang et al., 2009) and the > process of mental arithmetic (Anderson et al., 2008). But for the > cognitive scientists who use these methods, the primary question is > often not "how much" but rather "how" and "why" the patterns of neural > activity identified by a machine learning algorithm encode particular > cognitive processes. > > The aim of this workshop is therefore to 1) discuss the achievements > and problems of the decoding of high-level cognitive states, and 2) > explore the use of machine learning methodologies and other > computational models that enable such cognitive interpretation of > neural recordings of different modalities. Advances in this field > require close collaboration between machine learning experts, > neuroscientists and cognitive scientists. Thus, this workshop is > highly interdisciplinary and will aim to attract submissions also from > outside the existing NIPS community. By stimulating discussions among > experts in the different fields, the workshop seeks to generate novel > insights and new directions for research. > > Topics of interest > > The field requires techniques that are capable of taking advantage of > spatially distributed patterns in the brain, that are separated in > space but coordinated in their activity. Methods should also be > sensitive to the fine-grained temporal patterns of multiple processes > - which may proceed in a serial fashion, overlapping or in parallel > with each-other, or in multiple passes with bidirectional information > flows. Different recording modalities have distinctive advantages: > fMRI provides very fine millimetre-level localisation in the brain but > poor temporal resolution, while EEG and MEG have millisecond temporal > resolution at the cost of spatial resolution. Ideally machine learning > methods would be able to meaningfully combine complementary > information from these different neuroimaging techniques (see e.g. De > Martino et al., 2010). Moreover, as the processes underlying higher > cognition are so complex, methods should be able to disentangle even > tightly linked and confounded subprocesses. Finally, general use > algorithms that could induce latent dimensions from neural data, and > so reveal the "hidden" psychological states, would be a dramatic > advance on current hypothesis-driven analytical paradigms. Originality > of approach is encouraged and submissions on any related > methodological approach are welcomed, such as: > > - Interpreting spatial and temporal location of selected features and > their weights > - Discovering "hidden" or "latent" cognitive representations > - Disentangling confounded processes and representations > - Comparing or combining data from recording modalities (e.g. fMRI, > EEG, structural MRI, DTI, MEG, NIRS, EcOG, single cell recordings) > - Fuzzy and partial classifications > - Unaligned or incommensurate feature spaces and data representation > > As noted above, the complexity of higher cognition poses challenges. > To take language comprehension as an example, speech is received at > 3-4 words; acoustic, semantic and syntactic processing can occur in > parallel; and the form of underlying representations (sentence > structures, conceptual descriptions) remains controversial. We welcome > submissions dealing with any high-level cognitive functions that > exhibit similar complexity, for instance: > > - Knowledge representation and concepts > - Language and communication > - Understanding visual and auditory experience > - Memory and learning > - Reasoning and problem solving > - Decision making and executive control > > Submissions > > Authors are invited to submit full papers on original, unpublished > work in the topic area of this workshop via the NIPS 2011 submission > site at https://cmt.research.microsoft.com/NIPS2011/Default.aspx. > Submissions should be formatted using the NIPS 2011 stylefiles, with > blind review and not exceeding 8 pages plus an extra page for > references. Author and submission information can be found at > http://nips.cc/PaperInformation/AuthorSubmissionInstructions > . The stylefiles are available at > http://nips.cc/PaperInformation/StyleFiles > . Each submission will be reviewed at least by two members of the > programme committee. Accepted papers will be published in the workshop > proceedings. Dual submissions to the main NIPS 2011 conference and > this workshop are allowed; if you submit to the main session, indicate > this when you submit to the workshop. If your paper is accepted for > the main session, you should withdraw your paper from the workshop > upon notification by the main session. > > Important Dates > > - Aug 30, 2011: Call for papers > - Sep 23, 2011: Deadline for submission of workshop papers > - Oct 15, 2011: Notification of acceptance > - Oct 31, 2011: Camera-ready papers due > - Dec 16 or 17, 2011: Workshop date > > Links > > - NIPS 2011 website: http://nips.cc/Conferences/2011/ > - Workshop website: https://sites.google.com/site/decodehighcogstate > - Call for Papers: https://sites.google.com/site/decodehighcogstate/cfp/ > > Kind regards, > The Workshop Organizers, > Kai-min Kevin Chang, Anna Korhonen, Irina Simanova, Brian Murphy > > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: < > http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20110830/27b062df/attachment-0001.html > > > > ------------------------------ > > Message: 2 > Date: Tue, 30 Aug 2011 17:46:52 +0200 > From: Anna Lambrechts > To: fieldtrip at donders.ru.nl > Subject: [FieldTrip] Issues with neighbours in ft_timelockstatistics > Message-ID: > > > Content-Type: text/plain; charset="iso-8859-1" > > Dear fieldtrip community, > > Until about two weeks ago I was using a batch to process group comparison > statistics on MEG data using ft_timelockstatistics. A week later, the same > batch was not working anymore, even on the same data. > > The error I get is the following: > > *??? Improper index matrix reference. > > Error in ==> clusterstat>makechanneighbstructmat at 520 > [seld] = match_str(cfg.channel, cfg.neighbours(chan).label); > > Error in ==> clusterstat at 60 > channeighbstructmat = makechanneighbstructmat(cfg); > > Error in ==> statistics_montecarlo at 320 > [stat, cfg] = clusterstat(cfg, statrand, statobs,'issource',issource); > > Error in ==> statistics_wrapper at 290 > [stat, cfg] = statmethod(cfg, dat, design, 'issource',issource); > > Error in ==> ft_timelockstatistics at 124 > [stat, cfg] = statistics_wrapper(cfg, varargin{:}); > > Error in ==> myfunction_stats at 59 > gpstat = ft_timelockstatistics(cfg, data1, data2); > > *It seems the error concerns the 'neighbours.mat' file reading, even though > I am not sure about it. When I delete the two last lines in this file (to > match the size of matrix wanted) the error disappears, but the processing > is > incomplete and false (or at least it does not match the results I had > before). > > Has anyone encounter the same issue? > Did you do any update recently that might explain this problem, and do you > have any idea how to remedy to it? > > Best wishes, > Anna. > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: < > http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20110830/f81c8aad/attachment-0001.html > > > > ------------------------------ > > Message: 3 > Date: Tue, 30 Aug 2011 11:58:34 -0400 > From: Kanal Eliezer > To: Email discussion list for the FieldTrip project > > Subject: Re: [FieldTrip] assessing significance in using > ft_timelockanalysis results > Message-ID: <8CC6BA25-9A03-4630-B940-486BE2B9EE92 at cmu.edu> > Content-Type: text/plain; charset=us-ascii > > Hello Eric - > > Thanks for the response. It looks like the fixed vs random analysis is > exactly what I'm referring to. From what I understood, it looks like the > difference really only shows up in the variance of the resultant > distribution; with a random, the variance also takes into account the > betwee-subjects variance. Is there a way to specify whether I want to do a > fixed or random effects analysis in FieldTrip when I'm running > ft_timelockgrandaverage or ft_freqgrandaverage? Thanks! > > Elli > > p.s. - In case anyone else is trying to figure this out, chapter 12 of > Friston's book "Statistical Parametric Mapping" does an excellent job > explaining the difference between fixed and random analyses, as well as how > to implement it algorithmically. > > > > On Aug 26, 2011, at 3:00 PM, Eric Maris wrote: > > > Hi Kanal, > > > >> The event related statistics tutorial > >> (http://fieldtrip.fcdonders.nl/tutorial/eventrelatedstatistics) talks > > about > >> assessing significance parametrically by running t-tests on pooled > >> timelockanalysis data. My question is, does the fact that the averages > > were > >> created from N trials make a difference? If I'm condition A has twelve > >> averages and condition B has another twelve, and each average contains > 70 > >> trials, is there a way to "inform" the statistical test that the power > in > > this > >> dataset is greater than 24? Is this only possible if I run the t-test > > comparing > >> each set of 840 (70*12) trials? > >> > >> I'm also curious whether this is possible with non-parametric analyses, > as > >> well. Thanks - > > > > In an analysis over subjects (called random-effects analysis in the fMRI > > literature), "informing" the statistical test about the number of trials > per > > condition only makes sense if this number is different for the two > > conditions. I propose that you have a look the fMRI papers that deal with > > the issue of fixed-versus-random effect analyses. The conceptual issues > > involved are the same in fMRI and electrophysiology. > > > > > > Best, > > > > Eric Maris > > > > > > > > > >> > >> Elli Kanal > >> > >> > >> -------------------- > >> Eliezer Kanal, Ph.D. > >> Postdoctoral Fellow > >> Center for the Neural Basis of Cognition > >> Carnegie Mellon University > >> 4400 Fifth Ave, Suite 110A > >> Pittsburgh PA 15213 > >> P: 412-268-4115 > >> F: 412-268-5060 > >> > >> > >> _______________________________________________ > >> fieldtrip mailing list > >> fieldtrip at donders.ru.nl > >> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > _______________________________________________ > > fieldtrip mailing list > > fieldtrip at donders.ru.nl > > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > > > > ------------------------------ > > _______________________________________________ > fieldtrip mailing list > fieldtrip at donders.ru.nl > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip > > End of fieldtrip Digest, Vol 9, Issue 30 > **************************************** > -------------- next part -------------- An HTML attachment was scrubbed... URL: