Abstract
Echocardiography is a medical imaging modality that uses ultrasound in order to obtain cross sectional views of the heart. The basic problem in the use of echocardiography is the ability to obtain a reliable set of physical parameters related to cardiac status, so that assessment of heart disease can be performed automatically. This work overviews different image processing techniques used in the analysis of two dimensional echocardiographic images. After reviewing how the echocardiographic image formation process works, an outline of the general processing steps from image acquisition to automatic detection of important features is presented. Special emphasis on cardiac image segmentation is presented. In particular, a relaxation algorithm for image segmentation is discussed. Also, echocardiographic image segmentation using temporal analysis and a new algorithm for boundary detection is described. Measurements of left ventricular area, wall thickness, and ejection fraction is also presented. Shape analysis is introduced as a tool for echocardiographic image analysis. A high level description of the left ventricular boundaries using curvature is proposed. Curvature analysis attempts to identify stable landmarks during the beating process, muscles. Tracking these landmarks aids in the detection of abnormal heart contractions. Finally the use of expert systems is proposed in the analysis of echocardiographic images.
Date of this Version
7-1-1988
Comments
This work was partially supported by the "La Caixa" Program, Barcelona, Spain.