RAM-W600: A Multi-Task Wrist Dataset and Benchmark for Rheumatoid Arthritis

This is the benchmark code for the "RAM-W600: A Multi-Task Wrist Dataset and Benchmark for Rheumatoid Arthritis".
Dataset URL: https://huggingface.co/datasets/TokyoTechMagicYang/RAM-W600.

Update

• [2025-09-19] 🎉🎉Our dataset and benchmark paper has been accepted by NeurIPS 2025! 🎉🎉

• [2025-08-29] We conducted a major update of the RAM-W600 dataset.
  In this update, we expanded the dataset with 427 additional X-ray images, updated the split and released the metadata.

• [2025-05-10] We released the first update of the RAM-W600 dataset, which includes 621 X-ray images.

Setup

• Install the conda environment

conda create -n ramw600 python=3.10
conda activate ramw600

• Install PyTorch

# CUDA 12.6
pip3 install torch torchvision torchaudio

• Install other requirements

pip install -r requirements.txt

Dataset

Please refer to the link above to download the dataset.

Run

• Training
  The training configurations for segmentation and classification tasks are in ./train_seg.py and ./train_cls.py.
  We also provide scripts in ./train_seg.sh and ./train_cls.sh.
  Before running, you should refer to main_seg.py and main_cls.py and add your paths to the bash files.
  After running, the checkpoints will be saved in ./ckpts/.

bash train_seg.sh

• Testing
  The testing configurations for segmentation and classification tasks are in ./test_seg.py and ./test_cls.py.
  We also provide scripts in ./test_seg.sh and ./test_cls.sh.
  After running, the results of the visualization will be saved in the folder you chose for testing.

bash test_seg.sh

Citation

If you use RAM-W600 in your research, please cite:

@article{yang2026ram,
  title={RAM-W600: A Multi-Task Wrist Dataset and Benchmark for Rheumatoid Arthritis},
  author={Yang, Songxiao and Wang, Haolin and Fu, Yao and Tian, Ye and Kamishima, Tamotsu and Ikebe, Masayuki and Ou, Yafei and Okutomi, Masatoshi},
  journal={Advances in Neural Information Processing Systems},
  volume={38},
  year={2026}
}

Suggested Citation

If you use the benchmark code or experimental settings, we also recommend citing:

@misc{yang2026ramh1200,
  title={RAM-H1200: A Unified Evaluation and Dataset on Hand Radiographs for Rheumatoid Arthritis},
  author={Songxiao Yang and Haolin Wang and Yao Fu and Junmu Peng and Lin Fan and Hongruixuan Chen and Jian Song and Masayuki Ikebe and Shinya Takamaeda-Yamazaki and Masatoshi Okutomi and Tamotsu Kamishima and Yafei Ou},
  year={2026},
  eprint={2605.05616},
  archivePrefix={arXiv},
  primaryClass={cs.CV},
  url={https://arxiv.org/abs/2605.05616}
}