Automatic interpretation of echocardiograms-a computer vision approach

Cardiac function is often evaluated quantitatively using two-dimensional echocardiography to analyze shape attributes, such as the heart wall thickness or the shape change of the heart wall boundaries. A review of previous work in detecting the heart wall boundaries is presented, along with how this problem can be viewed from a computer vision perspective. The principles of echo image sequence analysis and high-level analysis are described. It is suggested that one promising approach is to use multiple-resolution processing by using a large window smoothed image for the initial detection of major edge segments, followed by smaller and smaller windows until a complete boundary is found.<<ETX>>