This repository contains the baseline model which reproduces the results proposed in the paper "Deep-learning-assisted diagnosis for knee magnetic resonance imaging: Development and retrospective validation of MRNet" by Bien, Rajpurkar et al.
Data must be downloaded from MRNet's official website and put anywhere into your machine. Then, edit the train_mrnet.sh script file by expliciting the full path to MRNet-v1.0 directory into the DATA_PATH variable.
To conduct adversarial training, please refers shell script: adv_train.sh with essential modification. If you meet problem running adversarial training, please contact me with email: [email protected]
The code was modified based on the GitHub Repository Matteo-dunnhofer/mrnet-baseline. Lines addition of the codes (only changed the original code here) in mrnet/train.py could be found in this report: /compare/diff.pdf. Besides, folder: experiment-bot and compare was designed and implemented with originality.
MIT