I'm trying to run TBSS on image data (http://users.loni.ucla.edu/~gprasad/images.zip), but the nonlinear registration step produces a strange transformation for image #2 (shrinking the image significantly). Do you know might be causing this issue? In addition, I have attached an example image #1 where the registration worked.
These nonlinear transformations were computed using fsl-4.1.5_64bit with tbss_2_reg -T.
The incorrect registrations cause the image mask in TBSS to be much smaller and therefore the skeleton to cover a smaller region.
|