Hi Martin,
so you think the current segmentation is worse than it used to be? Can you send us a dataset where this appears to be true? I wasn't sure from the images you sent which were old and which new, but they looked more like topology problems than segmentation ones (although the two are related)
cheers Bruce On Tue, 3 Mar 2009, Martin Ystad wrote:
Hello, I'm re-running some of my volumes previously processed through an older version of FreeSurfer (in 2004/2005 - version 3.01 I think). At the time they were processed through autorecon2 and 3, and problems that appeared were addressed by manual editing. However, when I re-process the volumes on version 4.2.0 in freesurfer, new segmentation errors appear, although I have kept my old manual edits from last time (see pictures). At the same time, we experience more than usual segmentation errors on new volumes that we process. Actually, many of these errors appeared after we updated from version 4.1 to 4.2. Is there any settings that I can tweak to make the segmentation algorithms behave more like they did before, or do I have to roll back to a previous version?
Thanks Martin Ystad