This repository contains the code for the paper "A network score-based metric to optimize the quality assurance of automatic target segmentation", currently under review.
You can reproduce the results of the article with the script his_main.py
- To get the score-based metrics, run cell 1.
- To get the correlation results from Tables 3 and S3, run cell 2.
- To plot figure 2, run cell 3.
- To get the AUC curves, run cell 4.
- To get scatter plots, run cell 5.
The nnU-Net was used as segmentation framework for this project (https://github.com/MIC-DKFZ/nnUNet). The trained models and JSON files related to its training are stored in the folder "training details".