/H-DenseUNet

TMI 2018. H-DenseUNet: Hybrid Densely Connected UNet for Liver and Tumor Segmentation from CT Volumes

Primary LanguagePython

H-DenseUNet: Hybrid Densely Connected UNet for Liver and Tumor Segmentation from CT Volumes, TMI 2018.

by Xiaomeng Li, Hao Chen, Xiaojuan Qi, Qi Dou, Chi-Wing Fu, Pheng-Ann Heng.

Introduction

This repository is for our TMI 2018 paper 'H-DenseUNet: Hybrid Densely Connected UNet for Liver and Tumor Segmentation from CT Volumes'.

Environment:

This code is only tested under python2. Check code environment "requirements.txt"

Usage

  1. Data preprocessing: Download dataset from: Liver Tumor Segmentation Challenge.
    Then put 131 training data with segmentation masks under "data/TrainingData/" and 70 test data under "data/TestData/".
    Run:

    python preprocessing.py 
  2. Test our model: Download liver mask from LiverMask and put them in the folder: 'livermask'.
    Download model from Model and put them in the folder: 'model'. run:

    python test.py
  3. Train 2D DenseUnet: First, you need to download the pretrained model from ImageNet Pretrained, extract it and put it in the folder 'model'. Then run:

    sh bash_train.sh
  4. Train H-DenseUnet: Load your trained model and run

    CUDA_VISIBLE_DEVICES='0' python train_hybrid.py -arch 3dpart
  5. Train H-DenseUnet in end-to-end way:

    CUDA_VISIBLE_DEVICES='0' python train_hybrid.py -arch end2end

Citation

If H-DenseUNet is useful for your research, please consider citing:

@article{li2018h,
title={H-denseunet: Hybrid densely connected unet for liver and tumor segmentation from ct volumes},
author={Li, Xiaomeng and Chen, Hao and Qi, Xiaojuan and Dou, Qi and Fu, Chi-Wing and Heng, Pheng-Ann},
journal={IEEE transactions on medical imaging},
volume={37},
number={12},
pages={2663--2674},
year={2018},
publisher={IEEE}
}

Questions

Please contact 'xmli@cse.cuhk.edu.hk'