Locor is a generic multimodal image registration tool based on modaling local functional dependence with linear combination of learned basis functions.
First, install PyTorch with instructions at https://pytorch.org/ (GPU version highly recommended).
Then you can install Locor using pip by running the command
pip install git+https://github.com/honkamj/locor
The registration is invoked via running python -m locor. The most basic use case is:
python -m locor <path_to_the_reference_image> <path_to_the_moving_image> -o <path_to_the_registered_moving_image.nii>
The images should be readable by nibabel (https://nipy.org/nibabel/). To see all the available options, run python -m locor -h.
Custom config files can be given with flag "-c". For Locor config files are defined as Python scripts defining a method "build_config". See the default config for an example.
If you use the method, please cite (see bibtex):
- New multimodal similarity measure for image registration via modeling local functional dependence with linear combination of learned basis functions
Joel Honkamaa, Pekka Marttinen
Under review (eprint arXiv:2503.05335)
Locor tool is released under the MIT license.