Skip to content

Latest commit

 

History

History
24 lines (16 loc) · 1.07 KB

README.md

File metadata and controls

24 lines (16 loc) · 1.07 KB

Source Matters: Source Dataset Impact on Model Robustness in Medical Imaging

This repository contains the code and results included in the paper (currently under review, link coming later).

Provided are the implementations of:

  • Confounder generation,
  • fine-tuning, and
  • performance analysis as described in the paper.

For access to the data used in the paper, please refer to:

Our dataset splits are provided in the data folder.

The main files to run are:

include_tag.py, include_lowpass.py, include_noise.py, gender.py (for confounder generation as described in the paper) fine-tuning.py (for fine-tuning the models on confounded targets and logging results) results_plots.ipynb and statistical_tests.ipynb (for pulling results together and plotting)

Feel free to contact us for help with the reproduction of our experiments.