what you are showing there is not a segmentation error but a severe misalignment. What is your freeview command line? Are you sure that the anatomicals and the surfaces are from the same subject?
On 7/10/2024 4:59 AM, Tornabene, Danilo wrote:
External Email - Use Caution
Dear FreeSurfer developers,
I've encountered very large, systematic errors in segmentation running recon-all on a MRI dataset of 76 subjects. In the screenshot attached, you can see how the white surface (shown in the right hemisphere) includes most of the cortex as well, while the pial surface (in the left hemisphere) extends into the CSF and often reaches the skull. In some of the subjects there is also a clear misalignment with the T1 images.
I don't have a long experience with FreeSurfer, but my supervisor Prof. Cole tells me he has processed thousands of scans and never seen this kind of errors so we thought of reaching out to you for some help or explanation. The only thing we noticed is that this error happens on scans with a higher-than-usual resolution (voxels are 0.8x0.8x0.8 mm), could this be causing the problem?
I have simply ran recon-all withouth any additional option or argument other than the input files and IDs and the output directory.
Please find the shell log outputs for two sample subjects. The recon-all pipeline was completed with no hard errors, and the Euler numbers were mostly 2 with some 1.
Best, Danilo Tornabene
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer