Dear Willeke,
Do you refer to coregistration of functional and structural data? The main reason for this preprocessing step is to align the two, which allows to apply normalisation parameters obtained from T1 segmentation onto the functional data. If you don't have any (good) T1 images then there's also nothing to segment, thus you would have to modify the preprocessing pipeline for some of the subjects (normalising the EPI means of the "bad" ones onto EPI template, possibly EPI segmentation). This is no good idea, within a single study preprocessing should be identical. The different routines are meant to normalise into MNI space, but one would expect some systematic differences nonetheless.
Best
Helmut
|