DEMS: Data-Efficient Medical Segmenter

Official Pytorch Implementation for Neural Networks Submission: “Segmenting Medical Images with Limited Data”

Preparation

Please ensure that a "checkpoint" and "dataset" folder exists in the project root directory following

├── project
    ├── data
    |   ├── busi
    |   |   ├── images
    |   |   ├── masks
    |   |   ├── indexes
    |   |   └── ...
    |   ├── ddti
    |   └── ...
    ├── src
    ├── checkpoint
    └── ...

To generate indexes for training and validation splits, execute

python split.py

Following this, you will be able to execute our DEMS for your unique application with

python train.py

Citation

If you find our DEMS useful for your research, please cite our paper as

@article{liu2024segmenting,
  title={Segmenting medical images with limited data},
  author={Liu, Zhaoshan and Lv, Qiujie and Lee, Chau Hung and Shen, Lei},
  journal={Neural Networks},
  pages={106367},
  year={2024},
  publisher={Elsevier}
}