As a quick update to this problem, it looks like the wm_mean and gm_mean are reasonable during the #SkullLTA and #SkullStripping steps. However, wm mean = 0 and gm mean = 0 for the #CA_normalize and #EM_registration steps.

Could this have anything to do with using a manual talairach alignment in previous steps? I did include the -nogcareg flag with the recon-all command.

On Mon, May 12, 2014 at 4:36 PM, Mark Plantz <markplantz2016@u.northwestern.edu> wrote:

Hi Freesurfers,

Does anyone know if it is possible to manually set the brain centroid during the EM Registration step of the recon-all pipeline. I am using a non-default atlas, and have discovered that many of my difficulties have resulted from the brain centroid being incorrectly assigned to a region of black space. Thank you for the help!

Best,

MP