/UPC-Pytorch

The code of ours paper "Uncertainty-aware Pseudo-label and Consistency for Semi-supervised Medical Image Segmentation"

Primary LanguagePython

UPC: Uncertainty-aware Pseudo-label and Consistency for Semi-supervised Medical Image Segmentationh

by Liyun Lu, Mengxiao Yin, Liyao Fu, Feng Yang.

Introduction

This repository is the Pytorch implementation of "Uncertainty-aware Pseudo-label and Consistency for Semi-supervised Medical Image Segmentationh"

Requirements

We implemented our experiment on the super parallel computer system of Guangxi University. The specific configuration is as follows:

  • Centos 7.4
  • NVIDIA Tesla V100 32G
  • Intel Xeon gold 6230 2.1G 20C processor

Some important required packages include:

  • CUDA 10.1
  • Pytorch == 1.6.0
  • Python == 3.8
  • Some basic python packages such as Numpy, Scikit-image, SimpleITK, Scipy ......

Usage

  1. Clone the repo:
git clone https://github.com/GXU-GMU-MICCAI/UPC-Pytorch.git 
cd UPC-Pytorch
  1. Download the Left Atrium dataset in Google drive. Put the data in './data/' folder
cd code/dataloaders
python la_heart_processing.py
  1. Train the model
cd code
python train_LA_UPC.py
  1. Test the model
python test_LA.py

Citation

Acknowledgement

Part of the code is revised from the UA-MT.

We thank Dr. Lequan Yu for their elegant and efficient code base.

Note