I was trying to create the white matter and pial surfaces of the cerebellum of the marmoset, based on ultra-high (80um) resolution ex-vivo images.

I manually segment the cerebellum into the white matter and the gray matter and change the header information to 1mm.

Based on the manual segmented files, I created white matter surfaces (rh.orig).

I tried to use the mris_make_surfaces to expand the white matter to create the matched pial surface.

However, the mris_make_surfaces failed to push the surface enough. See the attached files and screenshots.

I tried smoothing the segmentation, enhancing the contrasts, and tested different expert option (for example: -max_csf 0.1 -min_gray_at_csf_border 1)

This was this best I got, but still cannot get an optimal result (not pushed enough):

Since I have already manually segmented the cerebellum, is there a way to force the mris_make_surfaces to expand to the boundary of the image?

The files can be download to replicate what I did:

Thank you very much!