In fs5.3, I tried to generate a volume mask from the pial surface using mri_surf2vol: mri_surf2vol --mkmask --hemi lh --reg $subjsess/bold/001/register.dof6.dat --template $subjsess/bold/001/template.nii --surf pial --o $subjsess/bold/001/lh.pialmask.vol.nii.  The command run successfully.  I doubled check the registration between the EPI and high resolution T1.mgz under the $SUBJECTS_DIR, which has the mincost at 0.3898.  however, when I overlaid the mask on the T1.mgz, it is clear that some voxels outside brain are in the mask, and some voxels in the gray matter are not in the mask (see attached figure).  Did I use the function incorrectly?  

Thanks very much for your help.

Xiaomin