PyTorch implementation of the paper "EAGLE: An Edge-Aware Gradient Localization Enhanced Loss for CT Image Reconstruction". This repository includes the code for our novel Eagle-Loss function, designed to improve the sharpness of reconstructed CT image.

To ensure compatibility, please install the necessary packages using the following commands to create a conda environment and install eagle_loss package.:

git clone https://github.com/sypsyp97/Eagle_Loss.git
conda env create -f environment.yml
conda activate eagle_loss
cd Eagle_Loss
pip install -e .

FOV extension data can be downloaded here.

You can find the example usage in example.py.

Please cite the following paper and star this project if you use this repository in your research. Thank you!

@article{sun2024eagle,
  title={EAGLE: An Edge-Aware Gradient Localization Enhanced Loss for CT Image Reconstruction},
  author={Sun, Yipeng and Huang, Yixing and Schneider, Linda-Sophie and Thies, Mareike and Gu, Mingxuan and Mei, Siyuan and Bayer, Siming and Maier, Andreas},
  journal={arXiv preprint arXiv:2403.10695},
  year={2024}
}