Hi Matt

if they are outside of the pial surface at the end of recon-all then you don't have to worry about them.

cheers Bruce On Wed, 26 Feb 2014, Moore, Matthew wrote:

Dear Freesurfer Experts,

I am assessing the quality of the initial skull strip for some subject data after running autorecon1. I notice that in some cases non-brain matter near the eyes appears to remain, as well as some dura matter at the posterior of the brain. If these voxels are left untouched, or if some remain after either watershed adjustment or gcut, and autorecon2 is run, it appears that some of these voxels are identified as white matter (based on WM volume visualized in Freeview). I am wondering if the voxels are identified as white matter, but are not included in the surface reconstruction as shown by WM and pial surface outlines in Freeview, do they influence any of the measurement calculations (e.g. CorticalWhiteMatter) or segmentation accuracy downstream? In some troubleshooting pages I see examples that if they are included in the surface reconstruction, they are a problem, but if they are not, does the brainmask still require improvement? I am using Freesurfer 5.3 CentOS6 64bit. Thanks.

Matt