I am encountering a situation where some pial errors in a few subjects persist even after removing the non-brain voxels. It appears that pial surface has not changed although the portion in the brainmask has been erased. What could be some of the reasons for this? I have attached a screenshot.