Dear Chuyue
This issue goes away if you use the PEB framework. You analyse all the subjects together, take their parameters to the group level and including diagnosis as a regressor (covariate). Make sure not to bias the analysis by having different pre-processing procedures for each group.
Best
Peter
-----Original Message-----
From: Chuyue Zhao <[log in to unmask]>
Sent: Thursday, August 10, 2023 9:38 AM
To: [log in to unmask]; Zeidman, Peter <[log in to unmask]>
Subject: Re: Model selection and 2nd level analysis of DCM for patients
⚠ Caution: External sender
Dear Peter,
Thank you for the suggestion. I also have some following questions, as we want to compare the difference of effective connectivity between patients and health subjects, should we put data of patients and health subjects together for model selection or do the model selection seperately? Is it nessciary to maks the leison areas and using DTI results for using DTI as priors to inform the DCM connectivity?
Best,
Chuyue
|