This is the implementation of Cross-modal Memory Networks for Radiology Report Generation at ACL-IJCNLP-2021.

If you use or extend our work, please cite our paper at ACL-IJCNLP-2021.

@inproceedings{chen-acl-2021-r2gencmn,
    title = "Generating Radiology Reports via Memory-driven Transformer",
    author = "Chen, Zhihong and
      Shen, Yaling  and
      Song, Yan and
      Wan, Xiang",
    booktitle = "Proceedings of the Joint Conference of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing",
    month = aug,
    year = "2021",
}

• torch==1.5.1
• torchvision==0.6.1
• opencv-python==4.4.0.42

You can download the models we trained for each dataset from here.

We use two datasets (IU X-Ray and MIMIC-CXR) in our paper.

For IU X-Ray, you can download the dataset from here and then put the files in data/iu_xray.

For MIMIC-CXR, you can download the dataset from here and then put the files in data/mimic_cxr.

Run bash run_iu_xray.sh to train a model on the IU X-Ray data.

Run bash run_mimic_cxr.sh to train a model on the MIMIC-CXR data.