hmmm, can you upload the whole subject dir (tarred and gzipped) to our ftp server? It's probably a real anatomical feature, but it's impossible to say just looking at the images
On Thu, 10 Mar 2016, Selina Wilson wrote:
Hi Brian,
Thanks for your help. As you can see in the image, there are a couple places on this scan where the voxels are black and labeled "none". They appear throughout the brain, not just on this slice. Its as if a little (sometimes big) sphere of brain matter were removed at some point during the segmentation process or during the actual brain scan.
- Selina
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, March 10, 2016 11:26 AM To: Freesurfer support list Subject: Re: [Freesurfer] Subcortical segmentation voxel error
Hi Selina
can you post a picture showing the problem?
cheers Bruce
On Thu, 10 Mar 2016, Selina Wilson wrote:
Hello all,
I am new to Freesurfer and learning as I go/encounter problems. This might be an obvious question, but has anyone had issues with black holes in their subcortical segmentation runs? The wiki page it suggests contacting the freesurfer mailing list before attempting to manually fix the issue to see if there is an automatic procedure that could fix it.
The second part of my issue is that the black voxels are not incorrectly labeled as another structure/region. When I place the cursor over the black holes, the TkMedit Tools window reads "Sgmtn label: none" . If I do any of the suggested manual fixes (brushing, flood filling, linear selections) on the "none" voxels and label them as hippocampus, would that assign voxels to a structure, increasing the volume of that structure, when those voxels are not actually there?
To be more clear, I am worried that these "none" voxels are the result of movement during the scan or an error in how the scan was read by the computer. I do not want to 'fix' this problem by manually labeling those voxels as part of the hippocampus and find out down the line that this scan was actually unusable. That would mean that the data collected from this scan on hippocampal volume would be erroneous and make the final cross-sectional data analysis invalid/inaccurate.
Thank you for your help!
Selina Wilson Undergraduate Research Assistant Duke University Class of 2017 (317) 430 - 7404 selina.wilson@duke.edu