Compressed Sensing-Accelerated Cartesian Real-Time Phase Contrast MRI With Shared Velocity Encoding: A Clinical Performance Evaluation.

Abstract

Background: Conventional segmented phase contrast (PC) MRI is the clinical reference standard for cardiovascular hemodynamic quantification, but prolonged acquisition times, mandatory breath-holding, and susceptibility to arrhythmia limit its utility in vulnerable populations.

Purpose: To validate the clinical performance of compressed sensing-accelerated real-time (CSRT) PC MRI with shared velocity encoding (SVE) for fast, free-breathing cardiac flow assessment.

Study type: Prospective.

Population: 63 patients (46 male/17 female, mean age 51.5 ± 15.1 years) with diverse cardiac disease.

Field strength/sequence: 3.0 T; single-shot RF-spoiled gradient echo sequence with SVE (free-breathing CSRT PC MRI) and velocity encoded segmented gradient echo sequence (breath-hold routine segmented (RS) PC MRI).

Assessment: Hemodynamic parameters-peak velocity, peak flow rate, net flow rate, and peak mean velocity-were measured at the sinotubular junction plane for both CSRT and RS datasets. Scan time of the two sequences was recorded.

Statistical tests: Pearson/Spearman correlation analysis, intraclass correlation coefficient (ICC), Bland-Altman analysis, paired t-test/Wilcoxon test. A p value < 0.05 was considered significant.

Results: RS acquisition time was 16.0 ± 2.3 s; total imaging time including breath-hold instructions was 26.0 ± 2.3 s, whereas CSRT acquisition time was 10.0 s, yielding a statistically significant difference in scan duration. All hemodynamic parameters demonstrated strong significant correlation between CSRT PC and RS PC MRI (r ≥ 0.89) and good-to-excellent inter-method agreement (ICC ≥ 0.87). CSRT PC yielded significantly lower values for peak velocity (bias = -16.72 ± 11.48 cm/s [-13.31% ± 9.68%]), peak flow rate (bias = -36.28 ± 35.06 mL/s [-9.18% ± 8.98%]), and peak mean velocity (bias = -6.49 ± 7.06 cm/s [-11.02% ± 13.14%]) relative to the RS, but net flow rate was not significantly different between methods (bias = 0.15% ± 0.60 L/min [4.16% ± 17.28%, p = 0.058]).

Data conclusions: CSRT PC provided satisfactory agreement with conventional RS PC while shortening scan time and enabling free-breathing acquisition.

Evidence level: 1.

Technical efficacy: Stage 1.