Four-dimensional phase-contrast MRI with accelerated dual velocity encoding in patients with complex congenital heart disease

Aims
To evaluate, in complex congenital heart disease (CHD), the feasibility of a single free-breathing dual-VENC 4D-flow MRI acquiring low- and high-VENC data consecutively within the same acquisition, and to quantify how metrics from low vs. high VENC co-vary across vascular territories.

Methods and results
We conducted a single-centre, retrospective cohort study on a 3T system. Seventeen dual-VENC scans were acquired; 16 scans were analysable after quality review. We quantified the velocity-to-noise ratio (VNR), forward flow, peak velocity (Vmax), and the coefficient of variation (CV) of net flow across 30 cardiac frames in the ascending aorta (AAo), superior vena cava (SVC), inferior vena cava (IVC), pulmonary veins (PV), portal trunk (PT), and hepatic veins (HV). VNR was compared between VENC settings using a paired Wilcoxon signed-rank test. Agreement of forward flow and Vmax between low- and high-VENC was summarized with Spearman's rank correlation coefficient (r). Differences in variability (CV) were assessed with two-sided F-tests. Low-VENC increased VNR in AAo, PV, IVC, SVC, and PT (all P ≤ 0.003) and was not significant in HV (P = 0.54). CV was lower at low-VENC in AAo, PV, IVC, and SVC (all P ≤ 0.031). Agreement between low- and high-VENC was good-to-excellent in venous beds (SVC/IVC/PT r ≈ 0.67-0.88, all P ≤ 0.004), moderate in PV (r ≈ 0.57-0.62, P ≤ 0.024), but non-significant in AAo (r ≤ 0.55, P ≥ 0.17).

Conclusion
Dual-VENC 4D-flow MRI is feasible and clinically useful in complex CHD. Low-VENC improves VNR and reduces variability for slow venous flows, while high-VENC is required to faithfully capture arterial jets. Together in a single acquisition, they extend velocity dynamic range and streamline comprehensive haemodynamic assessment.