Detection of Boundaries of Carotid Arterial Wall by Analyzing Ultrasonic RF Signals

Abstract

In line with the fact that the intima–media thickness (IMT) of the carotid arterial wall is the most frequently used indicator to diagnose atherosclerosis by ultrasound, it is essential to accurately estimate the thickness of the intima–media complex (IMC) boundaries, i.e., the lumen–intima boundary (LIB) and media–adventitia boundary (MAB). In this study, an improved adaptive model of an ultrasonic echo was developed for the model to realize better fitting to the reference RF echo, which is measured from a glass plate, using a Gaussian window for the envelope function of the adaptive model. Using the mean squared error (MSE) method, the envelope of the improved adaptive model (multiply the sinusoidal wave with the Gaussian window) was fitted with the envelope of an RF echo measured in vivo to estimate the boundaries of the carotid arterial wall. Firstly, a computer simulation of the carotid arterial wall was conducted to evaluate the accuracy of boundary detection using the envelope of the improved adaptive model. In the simulation, IMT was set at 0.50 mm in a 7.2-mm-long short segment in the longitudinal direction. The IMT estimated by the proposed method was 0.54 mm. The 8% error between the true and detected IMTs showed the high accuracy of the envelope of the improved adaptive model in boundary detection. In the in vivo measurement, for the 4.8-mm-long short segment in the longitudinal direction, the average IMT automatically estimated by the proposed method was 0.57 mm. The result was compared with those obtained by the previous method and manually. The IMT estimated by the previous method, which uses an RF adaptive model, was 0.59 mm and the manually determined IMT was 0.56 mm. The smaller difference between the results obtained by the proposed method and manually verified that boundary detection by the proposed method was better than that by the previous method.