Skip to content

Latest commit

 

History

History
39 lines (28 loc) · 1.53 KB

README.md

File metadata and controls

39 lines (28 loc) · 1.53 KB

TandemNet

Deprecated. See new version at https://github.com/zizhaozhang/distill2

This is the implementation for the Oral paper titled "TandemNet: Distilling Knowledge from Medical Images Using Diagnostic Reports as Optional Semantic References", Zizhao Zhang et al, in MICCAI 2017. Please find more details in the paper.

Simple instructions of training the model

Prerequisites

The code is written in Torch7. Install necessary libraries:

Prepare training data

TandemNet takes images and corresponding text (diagnosic reports) as inputs in order to train. You need to write your own DataLoader.lua. An example with detailed explanations has been provided in utils/DataLoader.lua.

Training

sh scripts/train.sh

Evaluation

sh scripts/eval.sh

All results will be organized and saved in the folders inside checkpoints/tandemnet.

Please consider to cite the paper if it is useful

@inproceedings{Zhang2017TandemNet,
title={TandemNet: Distilling Knowledge from Medical Images Using Diagnostic Reports as 
Optional Semantic References},
author={Zhang, Zizhao and Chen, Pingjun and Sapkota, Manish and Yang, Lin},
booktitle={International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI)},
year={2017} 
}