Open-source MR imaging and reconstruction workflow.

Area of Science:

• Medical Imaging
• Computational Imaging
• Magnetic Resonance Imaging

Background:

• Magnetic Resonance (MR) imaging workflows often lack flexibility and interoperability.
• Vendor-specific tools hinder rapid prototyping and cross-platform collaboration.
• Reproducibility in MR imaging research is challenged by diverse acquisition and processing pipelines.

Purpose of the Study:

• To present an end-to-end, open-source MR imaging workflow.
• To enhance flexibility for rapid prototyping across the entire imaging process.
• To integrate vendor-independent, openly available tools for shared execution on different MR platforms.

Main Methods:

• Developed a workflow using Python and the Pulseq framework for MR sequence design.
• Integrated JEMRIS simulation framework for generating data from the same sequences used on scanners.
• Implemented a real-time image reconstruction and postprocessing pipeline using the Berkeley Advanced Reconstruction Toolbox.

Main Results:

• Demonstrated workflow flexibility with 3D parallel imaging (CAIPIRINHA), spiral imaging, and B0 mapping.
• Ensured all sequences, data, and processing pipelines are publicly available.
• Achieved real-time data processing as it is acquired.

Conclusions:

• The proposed workflow offers high flexibility, integrating advanced tools at all imaging stages.
• Open-source components simplify collaboration across MR platforms and sites.
• The workflow improves the reproducibility of MR imaging research results.