Recon-all fails with signal 11 during mris_fix_topology on FLAWS-hco data (giant defect, 240k vertices)

Software Support
1

Hi everyone,

I am trying to run recon-all on FLAWS-hco images (derived from a 3T FLAWS sequence, 1mm isotropic, with T1-like contrast: WM bright, GM gray, CSF dark). The pipeline successfully completes all steps up to topology fixing, but then crashes with signal 11 during left hemisphere mris_fix_topology.

FreeSurfer version: freesurfer-linux-ubuntu22_x86_64-7.4.1-20230614-7eb8460
OS: Ubuntu 24.04, kernel 6.17.0-14-generic
Command: recon-all -s subject01 -i FLAWS_HCO-DEN.nii.gz -all

**Error log:**CORRECTING DEFECT 0 (vertices=240980, convex hull=31271, v0=24)
Command terminated by signal 11

The defect size is enormous (240,980 vertices). I have tried:

  • Increasing swap space to ~15.6 GB total
  • ulimit -v unlimited
  • export OMP_NUM_THREADS=2
  • Memory was not exhausted during the run

Visual inspection confirms that wm.mgz contains mislabeled cerebellum and dura mater, which likely causes this giant defect.

Questions:

  1. Is manual editing of wm.mgz the only solution for such a large defect? If so, what is the recommended workflow in FreeView?
  2. Are there any flags or environment variables to make mris_fix_topology more tolerant to large defects?
  3. Has anyone successfully processed FLAWS-hco images with FreeSurfer?

Full log and screenshots attached. Any help would be greatly appreciated!

Thanks!