the problme is that you have several control points in voxels that aren't entirely wm. For example, 129, 134, 173 is a control point and it has no wm in it. This causes the intensity in that region to go up way too high, as we will normalize the control points to the desired wm intensity (110). I ran it without any control points and it worked pretty well. You should probably get rid of your control.dat. If you think that those thin frontal strands should go out a bit further, since some voxels that are entirely white matter (e.g. 134, 135, 162) but have an intensity that is less than 110 (this one is 103 after normalizating) and it will bring the intensity up a bit in that entire region (by the ratio of 110/103).
On Thu, 1 Dec 2016, Rizvi, Batool wrote:
Hi Bruce, Thanks for your reply. It is actually brighter on the brainmask.mgz than the orig.mgz when I checked. I'm uploading the freesurfer subject here.
Thanks! BR
From: freesurfer-bounces@nmr.mgh.harvard.edu [freesurfer-bounces@nmr.mgh.harvard.edu] on behalf of Bruce Fischl [fischl@nmr.mgh.harvard.edu] Sent: Thursday, December 01, 2016 2:31 PM To: Freesurfer support list Subject: Re: [Freesurfer] white matter segmentation incorrect after intensity change
hmmm, that's awfully bright. Is it also bright on the orig.mgz? It's not really possible to diagnose from a single slice from a single subject. If you tar, gzip and upload the subject one of us will take a look
cheers Bruce On Thu, 1 Dec 2016, Rizvi, Batool wrote:
Hi FreeSurfer experts,
After running -autorecon2 and -autorecon3, we're seeing issues for some of the subjects, which now start showing changes in intensity/brightness in some voxels, and this increased intensity is now missed by the white matter and grey matter segmentation, and is labeled as non-brain matter.
Attached is an example of a subject's brain that was segmented incorrectly, which we think is due to the intensity around that frontal region. We hadn't added control points in that region, so we are unsure what the cause of the intensity change is. In our first pass before running -autorecon2 -autorecon3, we did not notice this error or the intensity values to be so bright for that region.
Thanks for your help!
BR