> In order to get a more accurate segmentation result I segment my anatomical
> T1 images before coregistering them to the EPI images. Then, I coregister
> the grey matter T1 image to the unsegmented EPI image. The result looks
> fine. Are there any theoretical considerations why I should not coregister
> the segmented T1 image to the unsegmented EPI image?
This should work, but it isn't the optimal registration model to use for
registering tissue probability maps to images. The optimal strategy isn't
easily available within SPM5, but a few small modifications to spm_maff.m (or
a suitable wrapper function) would allow better coregistration.
Best regards,
-John
|