Validation of Left Ventricular High Frame Rate Echo-Particle Image Velocimetry against 4D Flow MRI in Patients

Abstract

AbstractBackgroundAccurately measuring blood flow patterns in the heart could provide insights in the pathophysiology of cardiac disease, and may provide additional diagnostic and prognostic information. This study aims to validate Echo-Particle Image Velocimetry (echoPIV) forin-vivoleft ventricular intracardiac flow imaging by comparing it with 4D flow MRI.MethodsWe acquired HFR contrast-enhanced ultrasound images from three standard apical views of 26 patients who required cardiac MRI. 4D flow MRI was obtained for each patient. Only echo image planes with sufficient quality and alignment with MRI were included for the validation. Regional velocity, kinetic energy and viscous energy loss were compared between modalities using normalized mean absolute error, cosine similarity and Bland-Altman analysis.ResultsAmong 24 included apical view acquisitions, we observed good correspondence between echoPIV and MRI regarding spatial flow patterns and vortex traces. The velocity profile at the cross-section at the base level (mitral valve) had cosine similarity of 0.92 ± 0.06 and normalized mean absolute error of 14 ± 5%. Peak of spatial mean velocity showed a difference of 3 ± 6 cm/s in systole and 6 ± 10 cm/s in diastole. The kinetic energy and rate of energy loss also revealed a high level of cosine similarity (0.89 ± 0.09 and 0.91 ± 0.06) with normalized mean absolute error of 23 ± 7% and 52 ± 16%.ConclusionsGiven good B-mode image quality, echoPIV can provide a reliable estimation of left ventricular flow, as compared with 4D flow MRI, providing comparable spatial-temporal velocity distributions. There are advantages and disadvantages for both modalities. EchoPIV captured inter-beat variability and had higher temporal resolution, while MRI was more robust to patient BMI and anatomy.CLINICAL PERSPECTIVEWe validated the results of high-frame-rate (HFR) 2D Echo-Particle Image Velocimetry (echoPIV) flow estimation against three-dimensional time-resolved phase-contrast cardiovascular magnetic resonance (4D flow MRI) imaging on three standard apical views of the left ventricle (LV) with sufficient B-mode image quality and plane correspondence.The flow results obtained with HFR 2D echoPIV showed good agreement with the velocities obtained with 2D projections of 4D flow MRI. We found that velocities and kinetic energy were comparable between the two modalities, but echoPIV measured higher viscous energy losses than 4D flow MRI. HFR echoPIV was able to capture intra-beat variability in flow patterns, which is not possible with 4D flow MRI, however the left ventricular outflow tract was often not captured in HFR echoPIV. Thus HFR echoPIV can be used to accurately assess flow in the LV and may provide an extra tool in the arsenal of cardiologists for the future study of cardiac disease.