Dear SPM experts,
I would need some help in setting the correct design matrix for my DCM analysis.
We are computing CMM_NMDA on TMS-EEG data. We have 2 groups of subjects doing 9 sessions of TMS-EEG over 24h. We computed the DCM over the 9 sessions (1 per session) for all subjects. Now I would like to test the effect of the session, and the interaction between session and group. I am not really sure how to do it, though. I have two main questions:
1) I have read on the SPM mailing list that you suggested a PEB of PEBs to someone in a similar situation. If I understood correctly, I should first define 2 separate PEBs for my 2 groups, regressing the effect of the session. Thus, in the GCM I would have 9 rows per subject? And how do I regress the session in the design matrix? Afterward, I should do a PEB of PEBs with group as covariate (1 for group 1, and -1 for group 2?)... But how do I perform a PEB or PEBs?
2) If I would like to use some parameters estimated in the DCM to compute others (as for instance a ratio), or if I would like to resample the parameters according to the subject's circadian phase, should I add the new parameters to the old DCM (or write a new DCM) before performing the session*group interaction (point 1)?
Thank you in advance for your time and help.
I am at your complete disposal if you need further information.
Kind regards,
Ilenia Paparella
|