Saner-deep-registration

TL,DR arXiv

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

• IXI
• OASIS
• BraTSReg

One can always use the google drive data files for IXI and OASIS datasets, kindly processed by Junyu Chen [here]. 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 [here].

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!