CXRFairness

Improving Fairness of Automated Chest X-ray Diagnosis by Contrastive Learning

Datasets

The first dataset is provided by Medical Imaging and Data Resource Center (MIDRC) and is available through this website (https://data.midrc.org/). The NIH-CXR dataset is available in this website (https://cloud.google.com/healthcare-api/docs/resources/public-datasets/nih-chest),

Getting started

Prerequisites

• python >=3.6
• pytorch = 1.11.0
• torchvision = 0.12.0
• sklearn
• pandas
• opencv
• skimage
• tqdm
• json
• pickle

Quickstart

I used the experiment on the MIMIC-CXR dataset on the sex groups based on the proposed method as an example.

python train_midrc_sex_supcon.py
then
python train_finetune_race.py

Reference

Acknowledgment

This work was supported by the National Library of Medicine under Award No. 4R00LM013001, NSF CAREER Award No. 2145640, and Amazon Research Award.