Dear Surfers,
I have a case where recon-all labels cerebellum as cerebral white matter (aparc+aseg.mgz) and the pial surfaces therefore include the cerebellum. The error has been resistant to attempts to fix the situation with manual edits to $SUBJECT/mri/brain.finalsurfs.mgz and re-running autorecon3 (with or without autorecon2).
Specifically:
Without manual edits, recon-all hangs at the stage: ... CORRECTING DEFECT 0 (vertices=34440, convex hull=821, v0=4192) XL defect detected... ...
After exiting with Ctrl-C, brain.finalsurfs.mgz include cerebellum, and the pial surfaces include cerebellum. I guess the "defect" FS is trying to correct is the cerebellum that was mislabeled.
My manual edits to brain.finalsurfs.mgz consisted of removing cerebellum by painting with brush value = 1 (freeview voxel edit tool). Thereafter, I recomputed autorecon 3:
recon-all -autorecon3 -s $SUBJECT -no-isrunning
With this, recon-all finishes "without error" but the recomputed aparc+aseg.mgz still thinks cerebellum is cerebral white matter and the pial surfaces again include cerebellum.
This is with Martinos Center stable version 6.
Any suggestions would be greatly appreciated! Also happy to send a link to the original T1 and recon-all results if you would like.
Bests,
Tommi
--- Tommi Raij, MD, PhD MGH/MIT Athinoula A. Martinos Center for Biomedical Imaging Bldg 149, 13th st Charlestown, MA 02129