Toward Fair and Accurate Cross-Domain Medical Image Segmentation: A VLM-Driven Active Domain Adaptation Paradigm

Thanks for your interest. Code is organizing and will be released.

How to Run the Code 🛠

Environment Installation

1. Training source models in the source center

For setting up the environment and training the source model, please refer to the [STDR] project. Please note that some hyperparameters, such as the image input resolution, may need to be adjusted.

2. Fair-AP strategy

3. Finetune the source Model with actively enhanced-pseudo labels

Please refer to the [STDR] project.

Citation 📖

If you find our work useful or relevant to your research, please consider citing:

@inproceedings{wang2025toward,
  title={Toward fair and accurate cross-domain medical image segmentation: A vlm-driven active domain adaptation paradigm},
  author={Wang, Hongqiu and Chen, Wu and Luo, Xiangde and Xing, Zhaohu and Liu, Lihao and Qin, Jing and Wu, Shaozhi and Zhu, Lei},
  booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
  pages={24102--24112},
  year={2025}
}

Comparison with Other Methods 📈

We acknowledge the developers of the comparative methods in ADA4MIA here.