MedMLX/swin-unetr-mlx

SwinUNETR · Abdominal CT

Segment abdominal organs in 3D CT with the MONAI BTCV checkpoint and native MLX.

GitHub · Files · Docs · Paper

Input	Output
3D abdominal CT · NIfTI	13 organ labels + background

Apple Silicon · Metal · Python 3.13

Quick start

uv venv --python 3.13
source .venv/bin/activate

uv pip install "swin-unetr-mlx>=0.2.6"
uv tool install huggingface-hub

hf download MedMLX/swin-unetr-mlx --revision 0052abb4bfb1c56bd7d52d5353267393399b8743 --local-dir models

Replace the example input paths with your own files.

from swin_unetr_mlx.medmlx import run

result = run(
    image_path='ct.nii.gz',
    weights_path='models/swinunetr.safetensors',
    output_dir='outputs',
)
print(result["outputs"])

Use & license

For research and education. Not validated for clinical use.

Weights: Apache-2.0. Notices.

Paper references

• Swin UNETR method: Self-Supervised Pre-Training of Swin Transformers for 3D Medical Image Analysis.

Citation

If you use MedMLX, please cite the MedMLX paper:

@article{sandoval2026medmlx,
  title   = {MedMLX: On-Device Medical Imaging AI for Apple Silicon},
  author  = {Sandoval, A. and Sandoval, J. and Saleem, A.},
  journal = {medRxiv},
  year    = {2026},
  doi     = {10.64898/2026.10.28.26366762}
}