Rethinking MRI as a measurement device through modular and portable pipelines

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Open Science

Background:

• Magnetic Resonance Imaging (MRI) faces challenges in reliable measurement due to proprietary systems and inconsistent implementations.
• This inconsistency leads to systematic variance in biomedical studies, hindering the clinical translation of quantitative imaging biomarkers (QIBs).

Purpose of the Study:

• To explore the benefits of open-source measurement pipelines in minimizing variability in MRI.
• To introduce a framework for benchmarking MRI workflows and a glossary for reproducibility strategies.

Main Methods:

• A tiered benchmarking framework (black-box to glass-box) was developed to assess measurement uncertainty propagation.
• Two modular and portable MRI measurement workflows were presented as practical implementation guides.
• A glossary of terms was provided to standardize strategies for reproducibility and post-hoc data pooling.

Main Results:

• The proposed framework exposes how workflow opacity contributes to measurement uncertainty.
• Illustrative workflows demonstrate how to decouple logic from computational processes for improved MRI pipelines.
• Standardized workflows can identify challenges in translating advanced frameworks to clinical settings.

Conclusions:

• Open-source MRI measurement pipelines are crucial for minimizing variability and establishing measurement uncertainties.
• Consistent terminology and standardized workflows are essential for enhancing the reproducibility and clinical translation of QIBs.
• Collaborative efforts among developers, regulators, industry, and clinicians are needed to advance glass-box MRI frameworks in clinical practice.