Code to the paper "Improving Automated Hemorrhage Detection in Sparse-view CT via U-Net-based Artifact Reduction" https://pubs.rsna.org/doi/10.1148/ryai.230275
How to use this code:
- Download the stage_2_train folder from https://www.kaggle.com/competitions/rsna-intracranial-hemorrhage-detection/data
- Unzip and place in ./Data/
- Create dataset with the construct_dataset.ipynb
- Train U-Net with the train_U_Net.ipynb
- Train EfficientNet with train_EfficientNet.ipynb
- Calculate ROC curves with eval.ipynb
- Calculate SSIM/PSNR values with Calculate_SSIM_PSNR.ipynb
The code was used on following libraries:
- tensorflow==2.4.0
- astra==2.1.0
- pydicom==2.3.0
- pandas==1.4.2
- sklearn==1.1.3
- skimage==0.19.3