Dear Yang
That's correct 😊
If you wanted to concatenate sessions, then it's straightforward to deal with head motion - you could either concatenate the head motion files vertically (stack one on top of the other) or you could enter them as separate regressors for each session. Trying both and seeing which gives you higher t-statistics for your task in the SPM analysis would help you to choose which is better.
Best
P
-----Original Message-----
From: SPM (Statistical Parametric Mapping) <[log in to unmask]> On Behalf Of Yang Xu
Sent: Wednesday, February 15, 2023 4:39 AM
To: [log in to unmask]
Subject: [SPM] DCM-PEB for multi sessions without concatenation
âš Caution: External sender
Hi all,
I hope to get some advice during the DCM-PEB analysis if I don't want to concatenate multi sessions.
In my project, each subject has six sessions (runs) with identical scans (333 scans of TR = 2s). According to some studies, I should concatenate the sessions, then extract the time series from VOI. However, I am unsure how to add head motion parameters to the first-level GLM. Say the data was preprocessed session-wise, and each session has its head motion parameters, if the six sessions were combined into one session, which head motion parameters should I add?
So I plan to create a DCM for each session and do the PEB-of-PEBs for group analysis. I want to ask your advice if I think correctly as below:
Generally, I will have a 3-level design, first-level: DCM for each run, second-level: PEB for each subject, third-level: PEB for the group.
GCM_subject1 = [DCM_sbuject1run1, DCM_sbuject1run2, ..., DCM_sbuject1run6];
PEB_subject1 = spm_dcm_peb(GCM_subject1, X1); X1 contains six rows of 1, as the mean of this group.
GCM_all = [PEB_subject1,..., PEB_sbuject50] PEB_all = [GCM_all, X2]; X2 contains one column of mean (i.e. 1) and other columns of covariates.
Do you think this approach is correct, or do you have other suggestions?
Best wishes,
Yang
|