Basically, I try to reconstruct a surface based on the average T1 images of a group of monkeys (see the attached figure). The original surface (?h.orig) looks fine, following the filled.mgz and wm.mgz very well. However, the wm and pial surfaces look messed up, not following the wm.mgz at all and the white matter surface and pial surface almost overlap on each other (i.e., almost zero thickness). I have reconstructed ~6 individual subjects' surfaces in which the T1 image was used to derive the template mentioned above and I have never seen this type of error.  I ran the recon-all twice and the results were almost identical. So, I wonder if anyone would tell me what is causing this and possible directions for avoiding such problems.

Many thanks in advance