Tutorial — first run¶

First run? Use this page, not the full CLI reference (same advice as DKT Connectome’s Tutorial).

1. Install tools¶

git clone https://github.com/phindagijimana/dlsurfer.git
cd dlsurfer
pip install -e .
source env/cluster_containers.sh   # points at lab SIFs + FS_LICENSE
dlsurfer doctor                    # expect [READY]

Details: Installation.

2. Dry-run a plan (no compute)¶

dlsurfer smoke
# or:
dlsurfer run --profile core-lesion \
  --t1 /path/T1.nii.gz \
  --lesion-mask /path/lesion.nii.gz \
  --sd /tmp/dlsurfer_tut --sid sub-01 \
  --fs-license $FS_LICENSE

You should see lit-inpainting → FastSurfer --seg_only → --surf_only → lit-post, with citations on each step.

3. Execute on a GPU node (optional)¶

dlsurfer run --profile core-lesion \
  --t1 /path/T1.nii.gz --lesion-mask /path/lesion.nii.gz \
  --sd ./subjects --sid sub-01 \
  --fs-license $FS_LICENSE --device cuda --execute

Or submit hpc/sbatch_eg_gpu_core_lesion.sh after editing paths.

4. Check outputs¶

subjects/sub-01/mri/aparc.DKTatlas+aseg.deep.mgz
subjects/sub-01/surf/lh.white
subjects/sub-01/scripts/dlsurfer_metrics.tsv

Layout: Outputs. QC notes: Quality control.

Public / lab evidence¶

EG (Gugger PTE) standalone FastSurfer 2.6 + LIT already completed on-cluster — EG smoke. TopoFit ±LIT machine score: TBI ± LIT report.