Awesome

Saner-deep-registration

TL,DR <a href="https://arxiv.org/pdf/2307.09696.pdf" target="_blank">arXiv</a>

We propose a novel regularization-based sanity-enforcer method that imposes two sanity checks on the deep registration model to reduce its inverse consistency errors and increase its discriminative power simultaneously.

Datasets

<ul> <li><a href="https://brain-development.org/ixi-dataset">IXI</a></li> <li><a href="https://learn2reg.grand-challenge.org/evaluation/task-3-validation/leaderboard">OASIS</a></li> <li><a href="https://www.med.upenn.edu/cbica/brats-reg-challenge"> BraTSReg</a></li> </ul> One can always use the google drive data files for IXI and OASIS datasets, kindly processed by Junyu Chen [<a href="https://github.com/junyuchen245/TransMorph_Transformer_for_Medical_Image_Registration">here</a>]. A big shout out to his effort.

Run

If data path is set, simply run sanity_checks_*.py

Trained Models

Trained models can be found [<a href="https://drive.google.com/drive/folders/1Ph_9T1Iw1YNy_13LgKxPC42mQm0Pxcda?usp=sharing">here</a>].

Bibtex

@inproceedings{duan2023sanity,
  title={Towards Saner Deep Image Registration},
  author={Duan, Bin and Zhong, Ming and Yan, Yan},
  booktitle={ICCV},
  year={2023}
}

Acknowledgment

This repo is heavily based on Junyu Chen's and Tony C. W. Mok's codes. Great thanks to them!