Dear Armin,

One simple way to do that would be to enter as your Gray/White/CSF ROIs just some meaningless mask (in surface-space), for example the one in ..../conn/utils/surf/mask.surface.brainmask.nii, and then in denoising simply make sure to list as potential confounders the correct White/CSF timeseries computed by fmriprep (e.g. from QC_a_comp_cor) instead of the ones computed by CONN (White/CSF). 

Hope this helps

Alfonso

Originally posted by Armin Toghi:

Dear Alfonso,

Thank you for your information on using fmriprep outputs for surface based analysis. I still have a problem which i cant figured it out. I would really appreciate your help.

I ran the fmriprep with --fsaverage flag, import struct as T1.mgz output, create func surface using conn_surf_gii2nii function and imported them to func files. However, when I want to go to the denoising step, I encounter this error 

"Attempting to extract BOLD signal from a surface-level file (AD_project/BIDS_test_prep/sub-001/func/func.surf.nii) using a volume-level ROI (AD_project/BIDS_test_prep/sourcedata/freesurfer/sub-001/mri/ec1T1.nii)."

It seems it tried to extract time series from gray matter mask that initially loaded after importing T1.mgz in structural file. I guess I should find a way to import WM GM CSF signal and ignore extraction from these volumes, since I just provided func surface. But I have no idea how. 

Thank you in advance for your help.

Best,

Armin