P727Left ventricular 2D flow pattern estimation by combining speckle tracking with Navier-Stokes based regularization: an experimental validation study


    loading  Checking for direct PDF access through Ovid

Abstract

Ultrasound imaging with low dose contrast injection would be of interest to better characterize left ventricular flow (LV) dynamics. We previously proposed a new estimation method by regularizing the initial motion estimates obtained through speckle tracking with the modified Navier-Stokes equations and tested this on simulated datasets. The aim of this study was to validate the method in an in-vitro ventricular flow phantom that reproduces flow conditions similar to the normal human heart but allows for simultaneous optical particle image velocimetry (OPIV) as a gold standard reference technique. An in-vitro left atria-ventricular simulator was built up with a bioprosthetic mitral valve. The mitral inflow profile (E/A = 1) and stroke volume (60 ml) were used as input to generate a 3D intra-ventricular flow field. Echo contrast (0.1ml Sonovue diluted into 50ml) was infused at 2ml/min and 2D B-mode images were recorded (frame rate: 178Hz; 45° sector angle). These echo data were processed using the proposed tracking methodology. The resulting 2D flow patterns were compared to the corresponding ground truth obtained by OPIV (double-pulsed mini Yag laser; CCD camera (Dantec Inc)). The histogram for the tracking amplitude error Δv = VTrack – VOPIV (normalized to the mean velocity) at all grid points was calculated and fitted by a normal distribution. An example of the estimated velocity field and the OPIV ground truth is shown in the figure. In panel b, the histogram for the tracking amplitude error is presented. The fitted normal distribution was centered at -23% with a standard deviation of 52.09%. These preliminary data show that the proposed tracking methodology underestimated the true velocity amplitude but that overall flow pattern characteristics were correctly visualized.

    loading  Loading Related Articles