iMED-Lab / UG-GAT

UG-GAT

This repository holds the Pytorch implementation of Uncertainty-guided Graph Attention Network for Parapneumonic Effusion Diagnosis (UG-GAT).

Introduction

We utilize all the CT images containing uncertainty information of a patient rather than a single 2D slice, and propose a graph-based framework for UPPE and CPPE classification.

Training BayesianCNN

BayesianCNN Training can be done:

python trainCNN.py

Obtaining feature and uncertainty for graph

After training the Bayesian, you can generate the image representations and uncertainty by running:

python test.py

Trainging UG-GAT

UG-GAT can be trained and tested by running:

python trainGraph.py

Citing This Paper

If you use this code,please use the following BibTeX entry.

  @article{hao2021uncertainty,
  title={Uncertainty-guided Graph Attention Network for Parapneumonic Effusion Diagnosis},
  author={Hao, Jinkui and Liu, Jiang and Pereira, Ella and Liu, Ri and Zhang, Jiong and Zhang, Yangfan and Yan, Kun and Gong, Yan and Zheng, Jianjun and Zhang, Jingfeng and others},
  journal={Medical Image Analysis},
  pages={102217},
  year={2021},
  publisher={Elsevier}
}