Automated Detection of Myocardial Wall Thickening for Echocardiographic Sequences

摘要

This paper proposes a method that stably detects myocardial wall thickening from echocardiographic images. Thickening of the myocardial wall is a useful index in the evaluation of cardiac function. In cardiac images obtained by the diagnostic ultrasound equipment, the epicardial contour is often obscure, and the accuracy is insufficient if the wall thickening is calculated from the endocardial and epicardial contours extracted from the edge information in the image. The proposed method is based first on the conventional method of automatic endocardium extraction [12], using an active contour model that takes account of the similarity of the image pattern near the contour. Then, the myocardial wall thickness parameters which maximize the similarity between the normalized images of the myocardial region and the template image for the myocardial region in the previous frame are sought, and the wall thickening is calculated. The normalized image is constructed by normalizing the shape of the myocardial region as determined by the myocardial wall parameters. Extraction of the epicardial contour from the edge information is not applied, and the deformation of the myocardium is detected by search in the myocardial wall parameter space, which helps to assure stable detection of the myocardial wall thickening. Evaluation using simulated images and real images demonstrates the effectiveness of the proposed method.