Establishing the reliability of non-contrast 4D flow in congenital heart disease at 1.5T

BACKGROUND: 4D flow has been shown to correlate well with 2D phase-contrast (2D PC) flow measurements, although most of the current literature is based on post-contrast imaging. OBJECTIVE: Given recent interest in non-contrast techniques, we hypothesized that non-contrast 4D flow (NC 4D Flow) would correlate well with 2D PC and ventricular stroke volumes. METHODS AND MATERIALS: A prospective cohort of pediatric and young adult patients with suspected/confirmed congenital heart disease (CHD) underwent cardiac MRI at 1.5T with acquisition of NC 4D Flow and 2D PC. Pearson correlation, linear mixed-effect correlations, and Bland-Altman comparisons were performed to compare these methods with secondary analyses of internal consistency and image quality/noise. RESULTS: In our cohort of 30 patients (age, 8-20 years), NC 4D Flow correlated strongly with 2D PC for net flow (r=0.76), peak velocity (r=0.9, P<0.001), and regurgitant fraction (r=0.75, P<0.001). By Bland-Altman comparison, NC 4D Flow was 10.1% lower for net flows and 18.9% higher for peak velocity compared to 2D PC. NC 4D Flow also slightly underestimated ventricular stroke volumes (18.1% lower for ascending aorta vs. left ventricular stroke volume and 13.9% lower for main pulmonary artery vs. right ventricular stroke volume). Both datasets were internally consistent; 2D PC had a higher relative signal intensity, but no difference in signal-to-noise or contrast-to-noise ratios (SNR/CNR). CONCLUSION: Non-contrast 4D flow correlates well with 2D PC in CHD with a slight underestimation in flows and overestimation in peak velocities, but good internal consistency and comparable SNR/CNR.