Dear Alfonso,

Does Conn implement a one voxel erosion of WM and CSF masks in aCompCor as stated in the Chai paper. I have subject that fails due to  the c3 ROI but not when I preform the segmentation externally in SPM and use the CSF mask generated there. I was wondering if this is the reason?

I suspect to do in SPM externally so as to include this subject would be incorrect if this is true?

Thanks for all your help.

Peter