Are you saying that synthseg itself is generating the wrong seg or that something is going wrong further downstream? The recon-all stream with synthseg is not a supported pipeline at the moment. For the standard stream, can you send the full recon-all.log file?

On 1/29/2024 2:04 PM, Maya Yablonski wrote:

External Email - Use Caution

Hello,

I am running freesurfer v7.3.2 reconall with -synthseg on a pediatric sample, and for one specific subject I keep getting a lot of gray matter classified as white matter, which leads to the ?h.white surfaces being too close to the pial surface. I have tried running with and without the --robust option but the issue recurs. Are there any expert options that I can play with before manually editing the WM.mgz? The issue is pervasive throughout the brain so will require heavily editing every single slice. The values for the misclassified gray matter are 85-97 in the brainmask.mgz and 100-110 in the WM.mgz and the T1 image looks ok by eye. When trying to run the same subject with the default pipeline (no synthseg), it crashes with the following error:

reading 1 input volumes... logging results to talairach.log reading '/share/software/user/open/freesurfer/7.3.2/average/RB_all_2020-01-02.gca'... GCAread took 0 minutes and 1 seconds. average std = 7.2   using min determinant for regularization = 5.2 0 singular and 884 ill-conditioned covariance matrices regularized reading 'nu.mgz'... error: mghRead(/MYPATH/freesurfer/sub-2341/mri/brainmask.mgz, -1): could not open file error: mri_em_register: could not open mask volume brainmask.mgz.

I wonder if there's any parameter I can tweak in the expert.opts? Thanks for any leads, Maya

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer