spatio-temporal-recon

Contact Joseph.Plummer@nih.gov/

Reconstruction code for handling 4D lung MRI data.

Created by Joseph Plummer, Ph.D. This code is built upon the fantastic and shareable code by the MoCoLoR group (Fei Tan, Ph.D., et. al.): https://onlinelibrary.wiley.com/doi/10.1002/mrm.29703.

Please cite them and this paper if you use the code.

Please post an issue on the repository page if there is a problem.

When making edits, please make a new development branch or make a fork.

Notice:

This is a work in progress repository.

Installation.

Run the following commands in sequence to run the experiments (note that we use Mamba instead of Conda, as it is faster).

1. mamba update -n base -c defaults mamba
2. make mamba
3. mamba activate spatio-temporal-recon
4. make pip

If you prefer to use Conda, replace all instances of mamba inside the Makefile with conda, and run the commands above with conda instead.

Data.

Requirements:

1. bksp = *.npy file with complex (binned) k-space data in form: [Nbins, Ncoils, Nexc, Nread].
2. bcoord = *.npy file with real (binned) k-space coordinates in shape: [Nbins, Nexc, Nread, 3] (note: scaled between +/- 0.5).

Troubleshooting:

1. This repository was tested on an NVIDIA GPU. If running on a system without the same, please remove the following packages from environment.yaml:
   • cudnn
   • nccl
   • cupy
2. Additionally, if not using an NVIDIA GPU, please set devnum = -1 inside the *.py files.

Running the scripts.

I run the shell script run_recons.sh to run the binning and reconstruction algorithms. Settings for each algorithm can be applied in this script.

Uninstall.

To uninstall, run the following commands:

1. mamba activate
2. make clean

DOI.

https://doi.org/10.1002/mrm.30416