[FieldTrip] Group level source statistics for individual condition

Mon Jul 18 16:28:38 CEST 2011

Hi Jan-Mathijs,

I see, this makes much more sense to me. So the anatomical information is preserved in the conduction model while the dipole grids are normalized, that makes sense. I'll give it a try and see what I get.

As for generating the second-level (group) statistics comparing baseline to activation, what would you recommend assuming I would like to submit t-statistics from individual subjects (comparing baseline to activation)?

Thank you so much for your help,
-Tony

From: fieldtrip-bounces at donders.ru.nl [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of jan-mathijs schoffelen
Sent: Monday, July 18, 2011 8:59 AM
To: Email discussion list for the FieldTrip project
Subject: Re: [FieldTrip] Group level source statistics for individual condition

Hi Tony,

Hmm, I always thought that it was better (more accurate at least) to perform the source estimation in a subject's native space prior to normalization. If I follow the steps outlined in that link then wouldn't I be performing the source estimation using a normalized grid based on the subject's mri instead of the native grid for that subject? It seems like there must be a way to perform the source estimation for each subject just like I normally do and then normalize those estimates prior to group level source statistics?

The idea is to still use subject specific information (i.e. the anatomical information per subject to build a volume conduction model of the head). Yet, for the definition of the single subject dipole grids, we will use a set of subject specific coordinates, which (after spatial normalisation) would yield the same coordinates for all subjects, thus bypassing the need to interpolate the functional data onto a high spatial resolution 3D grid. The trick is to 'unwarp' a set of canonical source coordinates into the subject specific head space.
In the second level step, ft_sourcestatistics can be tricked into accepting multiple subject data (each with unique source positions, as per their individual anatomy), if the source.pos per subject would be replaced with the set of canonical positions (which would be obtained after the spatial normalization).

Best,

Jan-Mathijs

Thank you,
-Tony

From: fieldtrip-bounces at donders.ru.nl<mailto:fieldtrip-bounces at donders.ru.nl> [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of jan-mathijs schoffelen
Sent: Monday, July 18, 2011 8:10 AM
To: Email discussion list for the FieldTrip project
Subject: Re: [FieldTrip] Group level source statistics for individual condition

Hi Tony,

We typically take a 2-step approach. First we compute per subject a descriptive statistic for a difference across 2 conditions (e.g. baseline vs activation, or two experimental conditions). This descriptive statistic can be a relative difference, absolute difference, or a T-statistic.
Then we typically do a second level statistical test, where we do statistical inference, using ft_sourcestatistics.

Note that in order to get your data into normalised space, you first have to call ft_sourceinterpolate, to create a data structure containing both the anatomy (which is to be spatially normalised in ft_volumenormalise) and the functional data.

Better still (at least computationally better) would be to create your single subject dipole grids (for the ft_sourceanalysis), based on a common space, as explained in the link attached.

http://fieldtrip.fcdonders.nl/example/create_single-subject_grids_in_individual_head_space_that_are_all_aligned_in_mni_space?s[]=mni&s[]=grid<http://fieldtrip.fcdonders.nl/example/create_single-subject_grids_in_individual_head_space_that_are_all_aligned_in_mni_space?s%5b%5d=mni&s%5b%5d=grid>

Best, Jan-Mathijs

I have tried to normalize the source results using volumenormalise in the past but I always received an error from sourcestatistics after performing that step, I think it has to be in the native source space to perform the statistics? I can run it again to find out the exact error that it gives.

Thank you,
-Tony

From: fieldtrip-bounces at donders.ru.nl<mailto:fieldtrip-bounces at donders.ru.nl> [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Rojas, Don
Sent: Friday, July 15, 2011 2:05 PM
To: Email discussion list for the FieldTrip project
Subject: Re: [FieldTrip] Group level source statistics for individual condition

Dear Tony,

You don't really give details of what you did on the single subject level that you want to try to achieve on the group level. But appending all the subject sources to each other does not make sense to me. Assuming you wrote out all the sources you computed as image files, then within Fieldtrip you can use ft_volumenormalise to write the results to standard source space (which uses SPM2 or SPM8). Or, you can just use SPM8 directly, taking the normalization result from the T1 MRI data and applying it to the source result computed within that space. That will put your subjects in MNI space and then you can do stats from there in a common anatomical space.

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
On Jul 15, 2011, at 11:04 AM, Passaro, Antony D wrote:

Hi everyone,

I tried to come up with a solution to this issue and the best I could do was to append all source analyses of pre-stim and post-stim trials for each subject to one another before performing source statistics across all trials and all subjects. However, I cant seem to find a way to normalize all subjects so they reside in the same source space prior to performing source statistics.

Any help would be much appreciated.

Thanks,
-Tony

From: fieldtrip-bounces at donders.ru.nl<mailto:fieldtrip-bounces at donders.ru.nl> [mailto:fieldtrip-bounces at donders.ru.nl] On Behalf Of Passaro, Antony D
Sent: Monday, July 11, 2011 8:25 AM
To: fieldtrip at donders.ru.nl<mailto:fieldtrip at donders.ru.nl>
Subject: [FieldTrip] Group level source statistics for individual condition

Hi all,

I'm trying to figure out how to compute the source statistics across a group of subjects for an individual condition (as compared to baseline). I have performed the single subject level source statistics for each condition compared to baseline (for all trials) and I have used those results to compute the group level statistics comparing both conditions directly. But I would also like to show the statistically significant sources across the group for each condition and while the average across subjects is informative, I would prefer to perform the same type of statistical analysis that was used to compare conditions directly. Is this possible with Fieldtrip and if so, would it require averaging the baseline and condition separately across all trials for each subject before comparing the condition and baseline at the group level using the sourcestatistics function?

Thank you for your help,
-Tony
_______________________________________________
fieldtrip mailing list
fieldtrip at donders.ru.nl<mailto:fieldtrip at donders.ru.nl>
http://mailman.science.ru.nl/mailman/listinfo/fieldtrip

_______________________________________________
fieldtrip mailing list
fieldtrip at donders.ru.nl<mailto: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<mailto:J.Schoffelen at donders.ru.nl>
Telephone: 0031-24-3614793

_______________________________________________
fieldtrip mailing list
fieldtrip at donders.ru.nl<mailto: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<mailto:J.Schoffelen at donders.ru.nl>
Telephone: 0031-24-3614793

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20110718/e0d421e1/attachment-0001.html>