Dixon-based B0 self-navigation in radial stack-of-stars multi-echo gradient echo imaging

Julia implementation of the Dixon B0 self-navigator based on the publication:

Jonathan Stelter, Kilian Weiss, Mingming Wu, Johannes Raspe, Philipp Braun, Christoph Zöllner, Dimitrios C. Karampinos; Dixon‐based B0 self‐navigation in radial stack‐of‐stars multi‐echo gradient echo imaging, Magnetic Resonance in Medicine, Magnetic Resonance in Medicine, DOI: 10.1002/mrm.30261, https://doi.org/10.1002/mrm.30261

The implementation of the B0 self-navigator can be found in src/Corrections/DeltaB0Correction.jl. Scripts for reproducing the simulations (with the exception of the commercial XCAT phantom) and phantom reconstruction are shared. The reconstruction was performed in Julia and the post-processing and evaluation in Python.

🚀 Setup

Requirements

• Julia 1.9 (system-wide installation is recommened)
• Anaconda/mamba environment with Python 3.10
• NVIDIA GPU recommended

Installing

1. Create a new mamba environment:

   mamba env create --name b0nav --file environment_nocuda.yml
   mamba activate b0nav
   which python

2. Open Julia and instantiate/precompile new Julia environment:

   using Pkg
   Pkg.activate(".")
   Pkg.instantiate()
   ENV["PYTHON"] = "/path/to/envs/b0nav/bin/python"
   Pkg.build("PyCall")

3. Run processing script directly from the shell:

   Replace n_threads with the number of threads you wish to use.

   julia -t n_threads -i scripts/2_1_recon_simulation_geo.jl

Data

Raw data for the simulation/phantom (>12GB) experiment are stored at the OneDrive folder.

Authors and acknowledgment

• Jonathan Stelter - Body Magnetic Resonance Research Group, TUM

Water-fat separation: https://github.com/BMRRgroup/fieldmapping-hmrGC

License

This project is licensed as given in the LICENSE file. However, used submodules / projects may be licensed differently. Please see the respective licenses.