High-Throughput Quantitative Chemical Shift-Encoded MRI of the Liver.
Abstract
Background
Access to MRI is limited by lengthy exam times and inefficient utilization. Focused protocols can reduce exam times, but workflow variability and inefficient room turnaround contribute to conservative scheduling with long exam slots.
Purpose
To develop and evaluate a high-throughput clinical MRI suite architecture and workflow, using an AI-prescribed free-breathing chemical shift-encoded (CSE) MRI exam to quantify liver proton density fat fraction (PDFF) in under 5 min of total MRI room time. STUDY TYPE Prospective. SUBJECTS 24 healthy volunteers in two cohorts: 12 research staff (7 women/5 men; age 26.8 ± 5.8 years) and 12 community volunteers (6 women/6 men; age 41.3 ± 13.5 years). FIELD STRENGTH/SEQUENCE 1.5 T; free-breathing 2D multi-echo gradient echo CSE-MRI. ASSESSMENT Each participant underwent three nonconsecutive CSE-MRI exams in a continuously queued workflow to characterize timing and PDFF repeatability. Workflow intervals were recorded from timestamped video review and image metadata. Staff cohort exams included two CSE-MRI acquisitions to assess within-exam repeatability, while community cohort exams included one to simulate clinical practice. Three radiologists (8/13/14 years of experience) independently evaluated AI-automated prescriptions for complete liver coverage and rated CSE-MRI image quality (five-point Likert scale). STATISTICAL TESTS Student's t-tests; Gwet's AC2; repeatability coefficients (RCs) with bootstrap 95% confidence intervals; Bland-Altman analysis. p < 0.05 was significant.
Results
Diagnostic image quality was achieved in all 72 exams (median PDFF Likert score 5/5, inter-rater AC2 ≥ 0.86). Total MRI room times averaged 4:09 ± 0:14 min (staff) and 3:35 ± 0:34 min (community). Turnaround times averaged under 2 min, enabling throughput of 16.1 exams per hour in the community cohort. Automated prescription achieved complete liver coverage in all exams. PDFF RCs were 0.78% (staff within-exam), 0.99% (staff between-exam), and 1.21% (community between-exam) absolute PDFF. DATA
Conclusion
The proposed high-throughput MRI workflow achieved over 16 exams per hour with highly repeatable liver fat quantification, demonstrating a framework for improving MRI utilization and access. EVIDENCE LEVEL 1. STAGE OF TECHNICAL EFFICACY 2.