MultiCore is a Python package for multi-contrast reconstruction of magnetic resonance images (MRIs). Powered by PyTorch, this library is designed to exploit parallel computing on multi-core processors -- whether in the form of CPUs or GPUs -- to maximally accelerate reconstruction time. MultiCore is also equipped with TensorBoard support, so that users may view images and track metrics as reconstruction takes place.
To install the software, clone this repository and run the following commands:
cd multicore-mri
pip install -e .
Then, you can run an example on the SRI 24 atlas with the following command:
python multicore/experiment.py dataset=atlas algorithm=bayesian_cs dataset.samp_type=y algorithm.alpha_init=1e3 algorithm.num_em_iters=15 algorithm.num_init_cg_iters=32 algorithm.num_probes=8 algorithm.alpha0=1e6 algorithm.grad_dim=xy dataset.samp_factor=0.249