Dear all,

For a longitudinal analysis I need to have within subject templates. After some preprocessing that included FSL's BET, and linear registration of all time points to time-point 1, I fed images to mri_robust_template:

mri_robust_template --mov tp1 tp2 tp3 --template <template_address> --satit

It runs fine for some subjects, however I get the following error for some others:

ERROR: All entries are zero! Images do not overlap (anymore?). This can have several reasons (i.e. different modalities, different intensity scales, large non-linearities, too diff. voxel sizes ...) Maybe use --transform <init.lta> with an approx. alignment obtained from tkregister or another registration program. Or do some prior intensity correction?

I have rechecked data, they are bias corrected, and coregistered.

Any thoughts of what could possibly go wrong here?

All the best, Arman