 |
 |
Dear Peter,
indeed, I though I ran the estimation over night but I actually did not....
a classical example of : "the solution is so simple and under your nose...but you still imagine it is a mountain to
climb" 🙂
Sorry for this stupid mistake!!
Best regards,
Claudia
Dear Claudia
First, are you certain that you estimated the group DCM (GCM) file for session 2 after running specify group?
If so, please could you check whether each DCM has identical data or different data? To do this, load the GCM file into the Matlab workspace by double clicking on it, then type or paste the code below.
Best
Peter
% Convert from DCM filenames to DCM structures if required
GCM = spm_dcm_load(GCM);
% Get subject 1's data
Y1 = GCM{1}.Y.y(:);
% Check for identical data across subjects
for subject = 2:size(GCM,1)
Y = GCM{subject,1}.Y.y(:);
assert(~all(Y == Y1), 'Subjects had identical data' );
end
% If we got this far, then all fine
disp('Done without errors');
From: Claudia Massaccesi <[log in to unmask]>
Sent: 06 December 2019 09:30
To: Zeidman, Peter <[log in to unmask]>; spm <[log in to unmask]>
Cc: Giorgia Silani <[log in to unmask]>
Subject: problem DCM for two sessions
Dear Peter,
I am trying to run two separate DCM analyses for the two sessions of my fMRI study (I decided to not concatenate the sessions because in between there was another task - manipulation - of ca. 20 min).
I followed the following procedure:
1) I created a DCM model for one subject for the first session using the GUI. I used the SPM matrix where I have both sessions (but not the manipulation task in between), however the GUI does not allow you to choose which session of the SPM matrix to use. Since I used VOIs from the first session, I imagined that the model was running on the first session only. (you can see my design matrix below).
2) I then replicated the model for all my subjects using the batch "Specify group" in SPM. I ran two times the analysis, one time specifying session 1 and adding the VOIs extracted from session 1; the second time specifying session 2 and adding the VOIs extracted from session 2. The SPM file and the DCM model where the same for both DCMs.
Now, the results from the first level DCM run on the first session look quite ok too me. However, the results from session 2 are very weird. The variance explained from the model is exactly the same for all subjects (19.90%) and when I try to open the diagnostic for each subject I get this error:
Error using plot
Vectors must be the same length.
Error in spm_dcm_fmri_check (line 272)
plot(t,DCM.y,t,DCM.y + DCM.R,':');
Error in spm_dcm_fmri_check>info_clicked (line 460)
spm_dcm_fmri_check(DCM, false, GCM);
Error while evaluating DestroyedObject Callback.
I guess I did something wrong. Do you have any idea? Shall I maybe first create two separated design matrix for session 1 and session 2 (by redoing all first levels), extract the VOIs and the run the DCM for each session?
Thank you in advance!
Best regards,
Claudia Massaccesi
The SPM matrix of one subject, the 1level DCM for the first session, the 1level DCM for the second session
-----------------------------------------------------------------
Claudia Massaccesi, PhD student
Faculty of Psychology
University of Vienna
Liebiggasse 5
1010 Vienna, Austria