ContiMorph: An Unsupervised Learning Framework for Cardiac Motion Tracking with Time-continuous Diffeomorphism
This repository is the official implementation of "ContiMorph: An Unsupervised Learning Framework for Cardiac Motion Tracking with Time-continuous Diffeomorphism".
- Please prepare an environment with python=3.9, run the following command
pip install -r requirements.txt
In our experiments, we used the following datasets:
-
3D cardiac MR images: ACDC dataset
-
3D cardiac MR images: M&Ms dataset
-
2D ultrasound images: CAMUS dataset
├─train ├─patient01 ├ ├─sequence1.nii.gz ├ ├─sequence2.nii.gz ├ └─... ├─patient01 ├ ├─sequence1.nii.gz ├ ├─sequence2.nii.gz ├ └─... ├─patient01 ├ ├─sequence1.nii.gz ├ ├─sequence2.nii.gz ├ └─... ├─... └─cine_files.txt ├─val ├─patient01 ├ ├─sequence1.nii.gz ├ ├─sequence2.nii.gz ├ └─... ├─patient01 ├ ├─sequence1.nii.gz ├ ├─sequence2.nii.gz ├ └─... ├─patient01 ├ ├─sequence1.nii.gz ├ ├─sequence2.nii.gz ├ └─... ├─... └─cine_files.txt ├─test ├─patient01 ├ ├─sequence1.nii.gz ├ ├─sequence2.nii.gz ├ └─... ├─patient01 ├ ├─sequence1.nii.gz ├ ├─sequence2.nii.gz ├ └─... ├─patient01 ├ ├─sequence1.nii.gz ├ ├─sequence2.nii.gz ├ └─... ├─... └─cine_files.txt
- To train the model run the following:
python ext/train_the_net.py
- To evaluate the model run the following:
python ext/test_the_net.py
model/pro_new_model.pth
Our code implementation borrows heavily from VoxelMorph,Deeptag and SGDIR.