Hi Dan
the problem is that you have artifacts above and below the brain that are very high intensity. mri_convert can deal with a few outlying values, but this is way too many for it. You could erase these manually, or use mri_extract to remove them.
cheers Bruce On Wed, 20 Jun 2018, Weisholtz, Daniel S.,M.D. wrote:
Hello all,
I am having a problem running recon-all with a particular subject due to a very large topological defect. I traced the problem to the white matter segmentation it it seems to be due to an inhomogeneity such that the WM is darker in the superior/anterior part of the brain and lighter in the inferior/posterior part of the brain. I was going to try to correct this problem with control points, but I discovered that starting with orig.mgz, all the brain volumes have a very reduced voxel intensity range (~166-168). rawavg.mgz has a much more typical-appearing voxel intensity range. So I think something is going wrong with the mri_convert --conform step that converts rawavg.mgz to orig.mgz. Since the control points seems to be incorporated at the autorecon2 stage, I’m not sure it would help here.
I am attaching rawavg.mgz and orig.mgz. I am very grateful for your thoughts.
Sincerely, Dan Weisholtz