SALAD: Self-supervised Aggregation Learning for Anomaly Detection on X-Rays

This repository contains the pytorch implementation of the proposed method SALAD: Self-supervised Aggregation Learning for Anomaly Detection on X-Rays which has been accepted for MICCAI 2020.

Citation

You find the PDF of SALAD: Self-supervised Aggregation Learning for Anomaly Detection on X-Rays MICCAI 2020 paper here.

If you use our code or find our work relevant to your research, please cite the paper as follows:

@inproceedings{bozorgtabar2020salad,
  title={Salad: Self-supervised aggregation learning for anomaly detection on x-rays},
  author={Bozorgtabar, Behzad and Mahapatra, Dwarikanath and Vray, Guillaume and Thiran, Jean-Philippe},
  booktitle={International Conference on Medical Image Computing and Computer-Assisted Intervention},
  pages={468--478},
  year={2020},
  organization={Springer}
}

Setup

Execute the following bash script:

bash setup.sh

It downloads the data and moves it to the data folder. It also creates a conda environment called SALAD, containing the necessary dependencies to run the code. Activate the conda environment with conda activate SALAD.

Train

Run the following command:

python main.py CXR_author unet ../log/salad ../data/author --w_contrast 0.25

Test

Run the following command:

python test.py CXR_author unet ../log/salad_test ../data/author --load_model ../log/salad/model_round10.tar

You can find our trained model checkpoints here.

License

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.