Algolzw / SPAC-Deformable-Registration

Reinforcement learning for deformable image registration, AAAI 2022

Official PyTorch implementation of the SPAC "Stochastic Planner-Actor-Critic for Unsupervised Deformable Image Registration", AAAI 2022.

Requirements

• Version

  • python 3.6
  • pytorch 1.4

• Install requirements:

pip install -r requirements.txt

Training

0. The core algorithm is in brain.py, and we build environment in env.py. The agent.py handles the logic of interactions.
1. Modify config.py to set path to data, iterations, and other parameters.
2. To train the model(s) in the paper, run this command:

python main.py

Evaluation

0. Set 'idx' parameter in config.py to choose existing models.
1. To test our model, run:

python test.py

Results

Citations

If our code helps your research or work, please consider citing our paper.

@inproceedings{luo2022stochastic,
  title={Stochastic Planner-Actor-Critic for Unsupervised Deformable Image Registration},
  author={Luo, Ziwei and Hu, Jing and Wang, Xin and Hu, Shu and Kong, Bin and Yin, Youbing and Song, Qi and Wu, Xi and Lyu, Siwei},
  booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
  volume={36},
  number={2},
  pages={1917--1925},
  year={2022}
}

@inproceedings{Luo2021StochasticAF,
  title={Stochastic Actor-Executor-Critic for Image-to-Image Translation},
  author={Ziwei Luo and Jing Hu and Xin Wang and Siwei Lyu and Bin Kong and Youbing Yin and Qi Song and Xi Wu},
  booktitle={IJCAI},
  year={2021}
}

If you have any question, contact me: [e-mail]