MRIReco.jl: An MRI reconstruction framework written in Julia

Area of Science:

• Medical Imaging
• Computational Science
• Software Development

Background:

• Magnetic Resonance Imaging (MRI) reconstruction demands high computational performance.
• Existing MRI reconstruction tools may lack flexibility or ease of use.
• Modern programming languages offer potential for improved performance and accessibility.

Purpose of the Study:

• Develop a high-performance, flexible, and user-friendly MRI reconstruction framework.
• Leverage the capabilities of the Julia programming language for scientific computing.
• Create an accessible platform for advanced MRI reconstruction algorithm development.

Main Methods:

• Utilized the Julia programming language for its efficiency in numerical computing and signal processing.
• Developed the MRIReco.jl framework, incorporating common imaging operators and MRI raw data format support.
• Integrated existing Julia packages to enhance modularity and functionality.

Main Results:

• MRIReco.jl provides a simple yet extensible interface for MRI reconstruction.
• The framework demonstrates comparable reconstruction speed to the established Berkeley Advanced Reconstruction Toolbox (BART).
• Julia's features facilitate efficient machine code generation, rivaling C/C++ performance.

Conclusions:

• Modern programming languages like Julia can bridge the gap between high performance and accessible scientific software.
• MRIReco.jl represents a promising environment for future advancements in MRI reconstruction algorithms.
• The framework offers a viable alternative for researchers seeking efficient and adaptable MRI reconstruction tools.