Hi, I don't know it's related, but we've seen something similar when rerunning recons initially processed in 4.05 in 4.1.0. Even though no initial manual edits had been made to the aseg, a file called aseg.maned.mgz (or something like that) was saved in the mri folder, and the medial/ventrical surfaces looked very much like yours. Adding the -clean-aseg flag to the recon-all command was the fix for us. Oddly, rerunning subjects from 4.01/4.03 -> 4.1.0 did not produce the same error.
Lars > We're experiencing some difficulties with editing several of our > control subjects' wm.mgz volumes. For many of them the ventricle > space was not filled during the recon-all process (some subjs have > enlarged ventricles, while others do not). Re-running with the - > bigventricles flag did not correct the problem, and we subsequently > filled the space manually by added new voxels (255 value) in the > wm.mgz volume. After running them again under -autorecon2-wm, we > discovered that the wm.mgz volumes now appear without any of our edits > (as if they had been run with a -clean-wm flag). We edited and reran > again, but with the exact same result for a second time, but only for > a subset of the original (edits worked on some, didn't on others). > Any thoughts as to why this is occurring? We are pretty confident > that the volumes are saved each time edits are made, and we are not > running with any -clean flags. If all else fails, is it permissible > to using the subjects as is (see attached pics) if the surfaces being > affected are only the 'unknown' regions? FS 4.1.0 is the version that > is being used, thanks. > > -Derin _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer