We provide the Keras implements based on Tensorflow Backend for REFUGE challenge segmentation task.

• python 3.5
• tensorflow 1.4.0
• keras 2.2.0
• GPU, CUDA

• tqdm
• skimage
• opencv
• scipy
• matplotlib

• Clone this repo:

git clone https://github.com/EmmaW8/pOSAL.git
cd pOSAL
python predict.py

python train_DGS.py
python test_DGS.py

Before running test, please check whether the model weight path is correct.

Acknowledge Some codes are revised according to selimsef/dsb2018_topcoders and HzFu/MNet_DeepCDR. Thank them very much.

@article{wang2019patch,
  journal={IEEE Transactions on Medical Imaging},
  title={Patch-Based Output Space Adversarial Learning for Joint Optic Disc and Cup Segmentation},
  author={Wang, Shujun and Yu, Lequan and Yang, Xin and Fu, Chi-Wing and Heng, Pheng-Ann},
  year={2019},
  volume={38},
  number={11},
  pages={2485-2495},
  publisher={IEEE},
  doi={10.1109/TMI.2019.2899910},
  }