Background

3D Magnetic Resonance Fingerprinting (3D-MRF) has demonstrated its ability to produce highly-reproducible quantitative parameter mapping from relatively short acquisitions. However, translating this technology into a clinically-feasible format capable of reconstruction within relevant timeframes remains a significant challenge. Previous MRF reconstructions have relied heavily on manual steps, impeding access to MRF maps for on-scanner image quality verification and timely delivery to radiologists. This delay in providing maps to clinicians hinders the incorporation of quantitative MRI into standard clinical workflows. To bridge this gap, we've developed and deployed a 6-minute whole-brain 3D-MRF protocol and automated reconstruction pipeline facilitating immediate access to quantitative T1 and T2 parameter maps, along with partial volume maps of white matter, gray matter, and CSF.
To our knowledge, this study represents the largest MRF investigation of the brain to date globally, aiming to establish the most diverse repository of clinical MRF brain imaging data. The comprehensive nature of this study not only demonstrates the robustness of our automated protocol and analysis pipeline but also positions online 3D-MRF as a benchmark for immediate availability of quantitative information in a format compatible with standard imaging practices.

Methods

A whole-brain 3D-MRF protocol was established with a field of view of 250x250x150mm³, spatial resolution of 1x1x2.5mm³, and total duration of 6 minutes. A custom analysis pipeline was designed and deployed within a secure cloud-based reconstruction cluster that seamlessly integrates reconstruction and registration utilities in a version-controlled and fully-traceable format. Notably, both the protocol and reconstruction pipeline require no manual intervention, allowing MRI technologists to perform MRF scans as part of all standard brain imaging workflows. The cloud-based reconstruction= accepts encrypted 3D-MRF raw datasets directly from scanners, generating tissue parameter and partial volume maps in both the patient coordinate space and registered to MNI-152 atlases for comparison against population values for target brain regions.

Results

The proposed online brain MRF acquisition and reconstruction has been successfully implemented across four MRI scanners at University Hospitals Cleveland Medical Center. With IRB approval, over 557 clinical patients have undergone examinations using this protocol since May 2023, with 97.6% of datasets reconstructed and returned to the scanner console while the patient is still in the bore (106±14 seconds). DICOM-compliant maps for all patients are available within PACS for research purposes, marking a substantial proof-of-concept integration of 3D-MRF into clinical workflows. Efforts are ongoing to extend this protocol to all clinical brain examinations at all sites within the UH system, aiming to establish quantitative baselines across diverse patient cohorts to inform future clinical decision-making.

Conclusion

This research holds significance in its direct integration with standard clinical procedures and in collecting one of the largest clinical quantitative multiparametric MRI datasets globally. The automated pipeline developed for 3D-MRF successfully offers near-immediate access to quantitative maps, enabling seamless integration into clinical workflows. This work represents a crucial step toward providing radiologists and clinicians with swift and reliable access to quantitative information, facilitating metric-driven clinical decision-making to improve patient outcomes.

Acknowledgements

No acknowledgement found.

References

[1] Ma, D., Gulani, V., Seiberlich, N., Liu, K., Sunshine, J. L., Duerk, J.L., & Griswold, M. A. (2013). Magnetic resonance fingerprinting. Nature, 495(7440), 187-192.