[FieldTrip] possible bug when using old neuromas data with newer fieldtrip version
russ port
russgport at gmail.com
Tue Jun 27 20:40:02 CEST 2017
Hi JM and fieldtriplist in general,
Sorry to bother you all again, but I think JM was correct; there is an ‘adder under the grass’.
Since the last email, I have re-read in my data using the newer version of fieldtrip.
Now the script no longer fails during artifact rejection! The issue now appears during ft_plot_sens. It is the same issue as last time, where due to the sens.tra code (ft_plot_sens -> lines 336,337):
poscoil = find(sens.tra(i,:)>0);
negcoil = find(sens.tra(i,:)<0);
returning more than one element for both poscoil and negcoil, fieldtrip then produces an error. I believe this failure is occurring because this neuromag data was collected with active shielding. As a result I had to put the data through MaxFilter before reading it into fieldtrip (as the built in fiedltrip warning suggests). Is this correct? Below is a link to the sens array
https://www.dropbox.com/s/y2n0iq5ew2g42f0/sens_tra_russ.mat?dl=0 <https://www.dropbox.com/s/y2n0iq5ew2g42f0/sens_tra_russ.mat?dl=0>
Does anyone have suggestions on how to proceed with a sens.tra that returns multiple positive and negative coils?
Best,
Russ
> On Jun 26, 2017, at 3:10 PM, Schoffelen, J.M. (Jan Mathijs) <jan.schoffelen at donders.ru.nl> wrote:
>
> Hey Russ,
> I saw your message come by on the discussion list. Apparently, nobody felt inclined yet to respond :(.
> Is there a reason why you read in the data with an old version of fieldtrip? I think there’s a backward compatibility clash, which is most likely due to the more fancy way in which sensors are defined these days. It could be that things are resolved if you upgrade your full analysis pipeline to a recent FT-version. Yet, I suspect an ‘adder under the grass’, as the Dutch saying goes…
>
> Best,
> JM
>
> PS: please do keep sending off questions to the list. The fact that nobody responds is usually not due to any lack of competence on your side, but is probably mainly due to everybody to be too busy with their own chores.
>
>
>> On 26 Jun 2017, at 20:32, Russell G Port <russgport at gmail.com <mailto:russgport at gmail.com>> wrote:
>>
>> Dear Dr. Schoffelen,
>>
>>
>> I was hoping to inquire about the email I sent last monday (see below). I have not seen it appear on the fieldtrip mailing list, and I am wondering if there was a problem with the email. I fully understand that if due to my past emails on the discussion list (and their lack of competence), my questions are no longer sent out. Alternatively, I was wondering if I should have sent this email to the bug tracker (though again I am not sure if it is actually a bug, or again my incompetence). I would really appreciate a confirmation, as I'm am currently held at my current step.
>>
>> Best,
>> Russ Port
>>
>> ---------- Forwarded message ----------
>> From: russ port <russgport at gmail.com <mailto:russgport at gmail.com>>
>> Date: Mon, Jun 19, 2017 at 9:54 PM
>> Subject: possible bug when using old neuromas data with newer fieldtrip version
>> To: FieldTrip discussion list <fieldtrip at science.ru.nl <mailto:fieldtrip at science.ru.nl>>
>>
>>
>> Hi All,
>>
>> Sorry to bother you all, but I was wondering if I may have found a possible bug. If so, I’ll submit a request to the bugzilla subsection. I am not completely sure the issue is not my own incompetence though. The error occurs when using a neuromag dataset that was read in with an old fieldtrip version (i.e. from 2015), and having it undergo ft_artifact_zvalue from a newer version of fieldtrip.
>>
>> In short, I use the following code:
>>
>> load('MEGGRAD_Visual_clean_combinedscripts')
>>
>> datanoline=[]
>> datanoline=data_clean
>>
>> load('/MEGGRADfinal_cleaned_jumpartifact_outputcfg_redone_Visual.mat')
>>
>>
>> cfgaj = [];
>>
>> % channel selection, cutoff and padding
>> cfgaj.artfctdef.zvalue.channel = allchanell;
>> cfgaj.artfctdef.zvalue.cutoff = cfgajout.artfctdef.zvalue.cutoff
>> cfgaj.artfctdef.zvalue.trlpadding = 0;
>> cfgaj.artfctdef.zvalue.artpadding = 0;
>> cfgaj.artfctdef.zvalue.fltpadding = 0.0;
>>
>> % algorithmic parameters
>> cfgaj.artfctdef.zvalue.cumulative = 'yes';
>> cfgaj.artfctdef.zvalue.medianfilter = 'yes';
>> cfgaj.artfctdef.zvalue.medianfiltord = 9;
>> cfgaj.artfctdef.zvalue.absdiff = 'yes';
>>
>> % cfgaj.artfctdef.zvalue.interactive = 'yes';
>>
>> [cfgajout, artifact_jump] = ft_artifact_zvalue(cfgaj,datanoline);
>>
>> When I run this code in a 2015 version of fieldtrip, it works fine (without any error). On the other hand, in the 2016 version of fieldtrip I have (as well as in the latest version which I downloaded this weekend), I get the following error:
>>
>> Error using ft_datatype_sens (line 278)
>> unexpected number of coils (408)
>> contributing to channel MEG0113 (1)
>>
>> Error in ft_datatype_raw (line 141)
>> data.grad =
>> ft_datatype_sens(data.grad);
>>
>> Error in ft_checkdata (line 236)
>> data = ft_datatype_raw(data,
>> 'hassampleinfo', hassampleinfo);
>>
>> Error in ft_artifact_zvalue (line 215)
>> data = ft_checkdata(data, 'datatype',
>> 'raw', 'hassampleinfo', 'yes’);
>>
>>
>> The dataset datanoline (as well as cfgaj) are linked to below (sorry it’s dropbox, but I’m not quite sure how to do it otherwise).
>>
>> https://www.dropbox.com/s/86t5agwv2vanrh3/MEGGRAD_Visual_clean_combinedscripts.mat?dl=0 <https://www.dropbox.com/s/86t5agwv2vanrh3/MEGGRAD_Visual_clean_combinedscripts.mat?dl=0>
>>
>> https://www.dropbox.com/s/rw8e7mof8plss72/MEGGRADfinal_cleaned_jumpartifact_outputcfg_redone_Visual.mat?dl=0 <https://www.dropbox.com/s/rw8e7mof8plss72/MEGGRADfinal_cleaned_jumpartifact_outputcfg_redone_Visual.mat?dl=0>
>>
>> Sorry for the bother,
>> Russ
>>
>>
>>
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20170627/f8fdadce/attachment-0002.html>
More information about the fieldtrip
mailing list