Deep Diffusion MRI Registration (DDMReg): A Deep Learning Method for Diffusion MRI Registration.
This code implements deep learning image registration method using diffusion MRI data, as described in the following paper:
Fan Zhang, William M Wells, and Lauren J O'Donnell. Deep Diffusion MRI Registration (DDMReg): A Deep Learning Method for Diffusion MRI Registration. IEEE TMI, 2022. [publisher's verion], [eprint], [supplementary results]
Installation:
conda create --name DDMREG python=3.8.5
conda activate DDMREG
pip3 install torch==1.9.0+cu111 torchvision==0.10.0+cu111 torchaudio==0.9.0 -f https://download.pytorch.org/whl/torch_stable.html
pip3 install joblib scikit-image pystrum nibabel
pip install jq
conda install -c mrtrix3 mrtrix
Example:
This exmaple shows a registraion of data from a teenager (~10 years old) to data from a healthy adult (~30 years old)).
git clone [email protected]:zhangfanmark/DDMReg.git
cd DDMReg
python ddmreg_register.py --modelDir ./ddmreg_models --movingDir ./test/sub_1/ --targetDir ./test/sub_2/ --outpuDir ./test/sub_1-TO-sub_2
Results:
After registraion, the mean Dice score across all tracts is 0.766. This value is 0.535 in the unregistered data.
Below is a visualization of the registration result. We suggest using 3D Slicer for a better visualiztion of the entire 3D FA volume.
Note:
- The testing data includes the direct input of the DDMReg network, i.e., FA and TOM images. The FA image is computed using SlicerDMRI, and the TOM images are computed using TractSeg.
- Preprocessing of the orignial images includes rigid registraion to the MNI space, upsampling to 1.25mm isotropic, and image padding to 128 × 160 × 128.
- The code for re-training a registraion model is under construction.