Dear CONN experts,

Do I need erosion of the GM/WM/CSF ROIs when I have imported them from external preprocessing?
I believe my segmentation in CAT12 should be very accurate. 


In my external pipeline, I did slice time correction and realignment in CONN, then I co-registered the structurals to the mean functional, then I segmented and normalised the structurals in CAT12, and used the deformation field from cat12 to normalise the functionals in SPM, before finally importing the output and smoothing it in CONN.

Still, I seem to get less connectivity results this way than using the default CONN preprocessing, and I don't think my initial results using purely CONN were noise, so I was wondering if I have some error with importing them.


Best,

Andreas