miccai17-mmwhs-hybrid

Thanks the great effort from Prof. Zhuang Xiahai in organizing the MMWHS Whole Heart Segmentation Challenge 2017. Please cite this paper if the code contributes to your work: "Hybrid Loss Guided Convolutional Networks for Whole Heart Parsing"
https://www.researchgate.net/publication/319702011_Hybrid_Loss_Guided_Convolutional_Networks_for_Whole_Heart_Parsing

you can find the required model below:
C3D Model:
https://drive.google.com/open?id=1N4LVb03Ehot34Bxcm1KkO14csuThOXUv
If you don't need this model, you can just comment the initialization of this model in code.

Trained model for CT segmentation:
https://drive.google.com/open?id=1Nly8ghHvedVC3EZetFJRRtLq7Ll-U0Fx

A CT data from the training dataset for demo:
https://drive.google.com/open?id=1b2sFaKrBfTRx6i0lBD5L6IZXP01W9L5C

If our work is helpful to you, please kindly cite our paper as:

@inproceedings{yang2017hybrid,  
title={Hybrid Loss Guided Convolutional Networks for Whole Heart Parsing},  
author={Yang, Xin and Bian, Cheng and Yu, Lequan and Ni, Dong and Heng, Pheng-Ann},  
booktitle={International Workshop on Statistical Atlases and Computational Models of the Heart},  
pages={215--223},  
year={2017},  
organization={Springer}  
}  

Segmentation Framework

Probability Maps Generated by Different Loss Functions (First row: weighted cross entropy. Second row: mDSC. Details in our paper)

Segmentation Results on CT and MR Volumes