TopoClassification

PyTorch implementation of the paper: 'PHG-Net: Persistent Homology Guided Medical Image Classification' (Accepted in the first round of WACV 2024).

This repo is still messy. I will make it more readable and provide more detailed docs. （too busy recently)

1. Data

download ISIC, Prostate, CBIS-DSM dataset and put them into the directory /home/data/raw_data

2. Train

run train.py. Gudhi package is utilized to generate the persistence diagram. There are many amazing tutorial on how to generate the persistence diagrams.

3. Citation

@article{peng2023phg,
  title={PHG-Net: Persistent Homology Guided Medical Image Classification},
  author={Peng, Yaopeng and Wang, Hongxiao and Sonka, Milan and Chen, Danny Z},
  journal={arXiv preprint arXiv:2311.17243},
  year={2023}
}