[Computerized processing of two-dimensional echo-cardiograms: its application for quantitating left ventricular regional contractility and three-dimensional echocardiography].

Abstract

This study assessed the computerized processing of two-dimensional echocardiograms for quantitating left ventricular regional contractility and for computer reconstruction of the left ventricle; so-called three-dimensional echocardiography. Computer analysis of two-dimensional echocardiograms for quantitative evaluations of the left ventricular regional contractility in myocardial infarction: A new practical method for computer digital image processing of two-dimensional echocardiograms was developed for quantitating left ventricular regional contractility. Short-axis cross-sectional images of the left ventricle at the levels of the mitral valve (MV), papillary muscles (PM) and the apex (AP) were recorded using a phased array sector scanner in thirty patients with healed myocardial infarction and fifteen normal control subjects. The echocardiographic data were recorded on the video tape and transferred to a minicomputer via interface circuits, digitized, and processed automatically. Each digitized image consisted of 256 X 256 pixels with a gray scale of 256 values. The edges of the endocardial and epicardial walls were detected by applying sequential steps including smoothing, dynamic thresholding, region growing, and filling of small holes. The rationale of edge detection depended on assuming that abrupt changes in gray levels occurred at the boundary, and that the points with maximum gradient values were boundaries of the left ventricular wall. Best-fit contours of endocardial and epicardial edges were drawn by fitting a spline. Besides this automatic edge tracing, endocardial and epicardial edges were also manually traced using a digi-pen system, whose results coincided well with automatic tracings. After detection of edges of endocardial and epicardial walls, the short-axis cross sectional left ventricular wall at each level was divided into eight segments. The geometric center of the end-diastolic left ventricular cavity and the axis connecting this with the posterior end of the right side of the septum was used as the reference point and line (fixed reference system). End-diastolic and end-systolic segmental hemiaxes, segmental area, segmental wall thickness and changes during the cardiac cycle were measured and calculated automatically in each segment using a computer. Regional contractility of the left ventricle was evaluated by percent systolic changes of the segmental hemiaxis, area and wall thickness. These values were significantly reduced in the infarcted left ventricular wall as defined by left ventriculography and coronary angiography.(ABSTRACT TRUNCATED AT 400 WORDS)