From Offline to Inline Without Pain: A Practical Framework for Translating Offline MR Reconstructions to Inline...

Area of Science:

• Medical Imaging
• Magnetic Resonance Imaging (MRI)
• Computational Imaging

Background:

• Offline magnetic resonance (MR) reconstruction methods face challenges during inline deployment.
• Common issues include scan disruption, limited multi-scan support, data format adaptation, and post-processing integration.

Purpose of the Study:

• To develop and validate a practical, open-source framework for inline deployment of established offline MR reconstruction techniques.
• To address scan disruption, multi-scan input limitations, data format variability, and scanner post-processing integration.

Main Methods:

• The framework utilizes the Gadgetron platform on Siemens scanners.
• Key features include an ISMRMRD to Siemens raw format converter, asynchronous trigger-and-retrieve, resource-aware scheduling, and integrated file management.
• Validation was performed on two Siemens scanners across SENSE, AlignedSENSE, and NUFFT reconstructions.

Main Results:

• The framework demonstrated minimal code modification for inline deployment, successfully executing reconstructions without disrupting scanner workflows.
• Automated and manual image retrieval was achieved, with scanner-based post-processing applied to custom outputs.
• Multi-sequence reconstructions were feasible for large-scale applications, with 99% of inline reconstructions retrieved successfully in 480 examinations.

Conclusions:

• The developed framework significantly reduces the technical barrier for inline deployment of offline MR reconstructions.
• It offers a robust, scalable, and post-processing-compatible solution for integrated MR imaging analysis.
• The open-source availability with documentation promotes reproducibility and community adoption.