This is the implementation of Cross-Modal Causal Intervention for Medical Report Generation. It contains the codes of the Radiological Cross-modal Alignment and Reconstruction Enhanced（RadCARE）, and fine-tuning via Visual-Linguistic Causal Intervention (VLCI) on IU-Xray/MIMIC-CXR dataset.

All the requirements are listed in the requirements.yaml file. Please use this command to create a new environment and activate it.

conda env create -f requirements.yaml
conda activate mrg

1. Datasets: You can download the dataset via data/datadownloader.py, or download from the repo of R2Gen. Then, unzip the files into data/iu_xray and data/mimic_cxr, respectively.
2. Models: We provide the well-trained models of VLCI for inference, and you can download from here.
3. Please remember to change the path of data and models in the config file (config/*.json).

• For VLCI on IU-Xray dataset

python main.py -c config/iu_xray/vlci.json

• For VLCI on MIMIC-CXR dataset

python main.py -c config/mimic_cxr/vlci.json

If you use this code for your research, please cite our paper.

@ARTICLE{11005686,
  author={Chen, Weixing and Liu, Yang and Wang, Ce and Zhu, Jiarui and Li, Guanbin and Liu, Cheng-Lin and Lin, Liang},
  journal={IEEE Transactions on Image Processing}, 
  title={Cross-Modal Causal Representation Learning for Radiology Report Generation}, 
  year={2025},
  volume={34},
  pages={2970-2985},
  doi={10.1109/TIP.2025.3568746}}

If you have any questions about this code, feel free to reach me ([email protected])

We thank R2Gen for their open source works.