Hi Nela,
This error has to do with the fact that your set of Neuromag sensors
is not full. There are 303 instead of 306. This might have to do
either with some channels being permanently broken on your system or
with something MaxFilter did. I suggest that you talk to Sven just to
be sure about the reason. In any case I fixed the code to be robust to
that. If you put the attached function in both
/external/fieldtrip/private and /external/fieldtrip/plotting/private
that should fix the problem.
Best,
Vladimir
On Wed, Apr 27, 2011 at 1:10 PM, Nela Cicmil <[log in to unmask]> wrote:
> Thank you. I include the .mat and .dat files in attached zip folder. Maybe the answer is that the preprocessing steps (MaxMove and MaxFilter by Neuromag) that I used, which removed the fiducial info, are incompatible with SPM 3D source reconstruction.
>
> Best,
> Nela
> ----
> Nela Cicmil
> D.Phil Candidate, Neurophysiology
> Dept. Physiology, Anatomy & Genetics
> University of Oxford
>
> Tel: 01865 282274
> Internal: 82274
> ________________________________________
> From: Vladimir Litvak [[log in to unmask]]
> Sent: 27 April 2011 09:44
> To: Nela Cicmil
> Subject: Re: [SPM] MEG forward model error 'subscript assignment dimension mismatch'
>
> OK. Send me just the mat file of that problematic dataset and I'll look at it.
>
> Vladimir
>
> On Wed, Apr 27, 2011 at 9:33 AM, Nela Cicmil <[log in to unmask]> wrote:
>> Dear Vladimir,
>>
>> Thank you very much for your reply. I have just checked that the bad channels were definitely marked in the epoched files before averaging and that they are consistent in the subsequent averaged files and grand-mean file that I am using for 3D reconstruction. I am trying to think of other possibilities: Maybe it is relevant that a previous preprocessing step (MaxFilter/MaxMove) had removed the fiducials from the 'raw' data, so I had to add the fiducials to the grandmean file by the command D = fiducials(D, fids); save(D), after copying 'fids' from D.fiducials of the relevant file that contained the fiducial information to which all sessions were aligned? Coregistration etc. looks fine on inspection.
>>
>> Also, I have noticed that in the data, I guess due to an error during data collection, between the sessions that are included in the grand mean, there are different channel 'types' associated with the channel numbers/labels: in some sessions the channels tab in the Display GUI shows:
>>
>> number label type
>>
>> 307 EOG061 ECG
>> 308 ECG063 EOG
>>
>> whilst in others:
>>
>> 307 EOG061 EOG
>> 308 ECG063 ECG
>>
>> It is the former description which is present in the grandmean file that I am using for source reconstruction. This hasn't been a problem so far as I did artefact removal on epoched files based upon channel 'type', which is correct wrt trace information, but could this label discrepancy somehow be causing the forward model to fail? If so, what is the correction that I should make to the converted data?
>>
>> Thank you for your help,
>>
>> Best wishes,
>> Nela
>>
>>
>> ----
>> Nela Cicmil
>> D.Phil Candidate, Neurophysiology
>> Dept. Physiology, Anatomy & Genetics
>> University of Oxford
>>
>> Tel: 01865 282274
>> Internal: 82274
>> ________________________________________
>> From: Vladimir Litvak [[log in to unmask]]
>> Sent: 27 April 2011 08:45
>> To: Nela Cicmil
>> Cc: [log in to unmask]
>> Subject: Re: [SPM] MEG forward model error 'subscript assignment dimension mismatch'
>>
>> Dear Nela,
>>
>> Might it be that you first prepared the head model and then marked
>> some channels as bad? In that case try deleting the source
>> reconstruction and starting again with the 'Template' button.
>>
>> Best,
>>
>> Vladimir
>>
>> On Tue, Apr 26, 2011 at 11:46 PM, Nela Cicmil <[log in to unmask]> wrote:
>>> Dear SPMers,
>>>
>>> I wonder if anyone could point out what might be going wrong to cause the following error message during Forward Model processing in spm8 for M/EEG:
>>>
>>> ??? Subscripted assignment dimension mismatch.
>>>
>>> Error in ==> channelposition at 147
>>> lab(i,:) = sens.label(sel)';
>>>
>>> Error in ==> ft_headmodelplot at 205
>>> [chan.pnt, chan.label] = channelposition(sens);
>>>
>>> Error in ==> spm_eeg_inv_checkforward at 74
>>> ft_headmodelplot(cfg);
>>>
>>> Error in ==> spm_eeg_inv_forward_ui at 44
>>> spm_eeg_inv_checkforward(D, val);
>>>
>>> Error in ==> spm_eeg_inv_imag_api>Forward_Callback at 87
>>> handles.D = spm_eeg_inv_forward_ui(handles.D);
>>>
>>> Error in ==> spm_eeg_inv_imag_api at 53
>>> feval(varargin{:}); % FEVAL switchyard
>>>
>>> ??? Error while evaluating uicontrol Callback
>>>
>>>
>>> I am performing a source localization analysis on a single subject using MEG data only from the Electa Neuromag scanner system, running spm8 (with r4290 updates) on matlab r2009a on mac 10.5.8. The data in question is epoched, has been grand-averaged, and contains fiducials. I specified the MRI individual head mesh using the subject's structural scan and co-registered using nasion, lpa & rpa: all of these steps ran without error and the resultant displays of brain, sensors, etc looked fine when I checked them.
>>>
>>> I originally tried forward model with the single shell head model, but it fails with the same error (above) regardless of which head model I use. The same error also occurs if I use the 'Template' instead of subject's own MRI.
>>>
>>> So, I was just wondering if someone could help tell me what this error means and if it is something that I must go back to fix in the analysis process? There might be some obvious error but I can't work out what it is.
>>>
>>> Many thanks for your help!
>>> Nela
>>>
>>> ----
>>> Nela Cicmil
>>> D.Phil Candidate, Neurophysiology
>>> Dept. Physiology, Anatomy & Genetics
>>> University of Oxford
>>>
>>>
>>
>
