Hello both John and Chris,
thanks for your replies.
First, Chris -
I am talking about FMRI data, and i've co-registered the EPI images (coreg: reference image=mean unwarped EPI, source image=T1 image).
Second, Chris and John -
If the problem lies in the co-reg (as Chris suggested) or in poor segmentation due to what John mentioned, how come using the segmented image (as opposed to the segmented parameter file) seems to work just fine?