[Freesurfer] repost - large-scale reconstruction errors

Hi,

In reconstructing some normal adult MPRAGEs, I've had a couple of different fairly large errors that I haven't been able to fix: (1) In a handful of subjects, more than half of the brain is lost, I think during the skullstripping step (ie, isn't present on the white matter or brainmask volumes later). This seems to be primarily on scans from a 3T Siemens with a 32channel head coil. I have tried adding control points, but this doesn't help. (2) In another handful of subjects, a large portion of the cerebellum is designated as cortex, and is labeled on the aseg maps.

Any suggestions on how to fix these two types of problems?

thanks, Nicole