As you mentioned it's an old data set and others might be running into similar problems, make sure that the EPI did actually cover the whole brain.
If the most anterior and dorsal parts were not within the field of view of the EPI sequence (but for the T1), then normalising EPI onto EPI template will warp the regions nearby outward, so that the EPI normalised volumes look as if you had covered the boundary areas. With T1 coreg and segmentation the cut-offs in the raw data would persist. The normalised EPIs would look odd then, but actually it would be the "correct" normalisation.
Best,
Helmut
|