DOI: 10.1002/jmri.70443 ISSN: 1053-1807

Autonomous Velocity Encoding ( VENC ) Selection Improves Precision of Quantitative Flow Measurement

Pierre Daudé, Rajiv Ramasawmy, Christine Mancini, Dominique Franson, Sunggun Lee, Anastasia Tsakirellis, W. Patricia Bandettini, Ahsan Javed, Kelvin Chow, Adrienne E. Campbell‐Washburn

ABSTRACT

Background

The optimal velocity encoding limit (VENC) in phase contrast MRI is subject‐specific because it depends on peak flow rate and the presence of flow jets. Currently, VENC is set manually with limited prior knowledge of the peak velocity. Setting the correct VENC might improve measurement precision or shorten scan times.

Purpose

To evaluate a workflow for autonomous selection of the optimal VENC without technologist interaction.

Study Type

Prospective and retrospective.

Population

The retrospective cohort included 254 scans from 113 patients (47 ± 14 years, 58 female) and 18 healthy volunteers (34 ± 15 years, 10 female). The prospective cohort included 5 patients with abnormal flow (47 ± 19 years, 2 female) and 10 healthy volunteers (28 ± 7 years, 6 female).

Field Strength/Sequence

1.5 T, 0.55 T; Cartesian gradient echo phase‐contrast sequence.

Assessment

A workflow has been designed to autonomously estimate and prescribe the optimal VENC. A 30s free‐breathing calibration scan was followed by automatic estimation of the maximum velocity (Vmax) in < 6 s. The calculated optimal VENC was then automatically applied in the subsequent flow measurement without operator intervention. The target VENC:Vmax ratio was 1.1–1.25 according to consensus statements.

Statistical Tests

Shapiro–Wilk test, nonparametric two‐sided bootstrap with 90% confidence intervals, Wilcoxon signed‐rank test, paired t ‐test. Holm correction was applied to prespecified pairwise comparisons. p  < 0.05 was considered significant.

Results

Retrospective analysis demonstrated sub‐optimal VENC setting in 96.5% of examinations with a VENC:Vmax ratio [2.03, 2.13] (two‐sided bootstrap 90% confidence interval). In the prospective cohort, autonomous inline VENC selection yielded a mean VENC:Vmax ratio of 1.19 ± 0.10, significantly lower than subject‐invariant VENC settings (VENC 150 :Vmax = 1.49 ± 0.31, VENC 200 :Vmax = 1.99 ± 0.41). Optimized VENC improved measurement precision, reducing velocity standard deviation by 41.9% ± 8.0%, and enabling shorter scan time to approximately 1/3 compared with default VENC 200 cm/s while maintaining equivalent velocity‐to‐noise performance.

Data Conclusion

Inline autonomous VENC selection improved flow‐measurement precision, simplified the acquisition workflow, and reduced scan time at 0.55 T.

Evidence Level

2.

Technical Efficacy

1.

More from our Archive