Dear all,
I am trying to make group level analysis of functional data on the surface and add per-vertex regressors (PVR, here: cortical thickness) into the design. When I view the results I have the feeling that the PVR is not really included in the model:
I am using the command:
mri_glmfit --glmdir lh_output --y cope1_lh_4D.mgh --fsgd Group1versusGroup2.fsgd --pvr lh.thickness.10.mgh --C Group1only_thickness.mat --C Group2only_thickness.mat --C Group1minusGroup2_thickness.mat --C Group2minusGroup1_thickness.mat --C Group1plusGroup2_thickness.mat
where
cope1_lh_4D.mgh contains the concatenated sig.mgh images from the fixed level analysis (combining the runs) for each participant
fsgd file contains:
GroupDescriptorFile 1
Title Group1versusGroup2
Class Group1
Class Group2
Input 1109 Group1
Input 1110 Group1
Input 1111 Group1
Input 1112 Group1
...
Input 2280 Group2
Input 2282 Group2
Input 2283 Group2
Input 2284 Group2
The PVR is a file containing thickness maps for each participant, concatenated in the same order as for the functional data.
So there are 2 main regressors (1 factor, 2 levels), but the contrast files contain 3 regressors (the third one is the PVE), for instance:
Group1minusGroup2_thickness.mat contains 1 -1 0.
Does it look correct?
Should the pvr regressor be included in the fsdg file? How can I do it?
Thank you very much for your help,
Cheers,
Aga