P1C-3 Correlation Estimator Compounding for Reduction of Variance in Elasticity Imaging of Arterial Wall

Abstract

Pathological changes in arterial walls significantly influence their mechanical properties. We have developed a correlation-based method, the phased tracking method, for measuring the regional elasticity of the arterial wall. Using this method, elasticity distributions of lipids, blood clots, fibrous tissue, and calcified tissue have already been measured in vitro. However, these elasticity distributions were found to overlap each other, and this overlap worsens the tissue classification based on elasticity images. In this study, spatial compounding of the correlation estimator for estimating strain was introduced to reduce undesirable variances in elasticity distributions. To determine radial strain (change in thickness) of the arterial wall caused by heartbeat, a combination of two points are assigned along an ultrasonic beam, and the rate of the change in thickness between these two points (=layer) are estimated. Then, the combination of two points is slid along the ultrasonic beam with an interval of sampled points to obtain the spatial distribution. In this process, the initial distance between two points is set to be larger than the duration of ultrasonic pulse (eight times of the interval of sampled points), and the combination of two points is slid with an interval of sampled points. Therefore, layers, which are defined by the region between each combination of two points, overlap each other. In this study, complex correlation functions of overlapping layers are compounded to reduce the variance in estimated rates of changes in thickness. Temporal integration is applied to the estimated rate of change in thickness to obtain the radial strain and elastic modulus. Eighteen arteries (femoral and iliac) were measured in vitro. The change in internal pressure was applied using a flow pump. By comparing measured elasticity images with corresponding pathological images, elasticity distributions of lipids, blood clots, fibrous tissue, and calcified tissue (meanplusmnSD) were determined to be 67plusmn37, 105plusmn76, 989plusmn1359, and 1292plusmn839 kPa, respectively, by the previous phased tracking method. By the proposed method with spatial compounding, they were determined to be 89plusmn47, 131plusmn56, 1022plusmn1040, and 2267plusmn1228 kPa, respectively. The proposed spatial compounding suppressed variances and increased the difference among their mean elasticity. This result shows that the proposed method improves the tissue classification based on elasticity