DiffuseMorph

This repository is the official implementation of "DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model" which is presented at ECCV 2022.

[ECCV paper]

Requirements

OS : Ubuntu
Python >= 3.6
PyTorch >= 1.4.0

Data

In our experiments, we used the following datasets:

2D facial expression images: RaFD dataset
3D cardiac MR images: ACDC dataset
3D brain MR images: OASIS-3 dataset

Training

To train our model for 2D image registration, run this command:

python3 main_2D.py -p train -c config/diffuseMorph_train_2D.json

To train our model for 3D image registration, run this command:

python3 main_3D.py -p train -c config/diffuseMorph_train_3D.json

Test

To test the trained our model for 2D image registration, run:

python3 main_2D.py -p test -c config/diffuseMorph_test_2D.json

To test the trained our model for 3D image registration, run:

python3 main_3D.py -p test -c config/diffuseMorph_test_3D.json

Citation

@inproceedings{kim2022diffusemorph,
  title={DiffuseMorph: Unsupervised Deformable Image Registration Using Diffusion Model},
  author={Kim, Boah and Han, Inhwa and Ye, Jong Chul},
  booktitle={European Conference on Computer Vision},
  pages={347--364},
  year={2022},
  organization={Springer}
}

DiffuseMorph / DiffuseMorph