Some comment to my last posting, as John said, it might really depend on the data. I should really have stressed that unified segmentation lead to a better overall normalization >>in my very specific data sets<<.
If you want to make sure, just apply the different normalization parameters to your structural volumes, and create mean and SD volumes across the different subjects for n. based on segmentation, n. based on EPI template, n. based on T1 template. In my case, e.g. the amygdala and the POS were blurred to a lesser extent with segmentation compared to T1 normalization.
|