SCALA checkpoints
Pretrained weights for SCALA: Semi-supervised Cascade for Left Atrial Scar, Cavity, and Multi-Structure CT Segmentation (MICCAI 2026 CARE LeftAtrium challenge).
Code and usage: https://github.com/adinathdukre/SCALA
| Folder | Task | Trainer |
|---|---|---|
cavity_resenc |
LA cavity (Task 2), cavity localizer for scar (Task 1) | nnUNetTrainer (ResEnc-L) |
cavity_mednext |
LA cavity (Task 2) | nnUNetTrainerMedNeXt |
scar_resenc |
LA scar (Task 1) | nnUNetTrainer (ResEnc-L) |
scar_surface |
LA scar (Task 1) | nnUNetTrainerScarSurface |
ct_resenc |
CT LA / LAA / PV (Task 3) | nnUNetTrainer (ResEnc-L) |
ct_stunet |
CT LA / LAA / PV (Task 3) | STUNetTrainer_base_ft |
Each folder is an nnU-Net v2 result folder: plans.json, dataset.json, and fold_0 to fold_4, each holding checkpoint_best.pth. Optimizer state has been removed.
huggingface-cli download adidukre/SCALA --local-dir /path/to/checkpoints
python -m scala.predict --task task1 -i /path/to/input -o /path/to/output -m /path/to/checkpoints