COMPARISON OF THREE DIFFERENT METHODS FOR AUTOMATED DISCRIMINATION OF MYOCARDIAL HEART DISEASE

D.-Y. Tsai, Y. Usui, K. Kojima

Abstract

The aim of this paper is to compare the performance of three different methods, i.e., neural network with backpropagation learning, neural network with genetic-algorithm-based learning, and genetic-algorithm-based (GA-based) fuzzy logic approach, for automated discrimination of myocardial heart disease. In our experiments, a total of 90 samples of echocardiographic images from 45 subjects were used. Four statistical features, namely, angular second moment, contrast, correlation and entropy, were extracted from each image. These four features were subsequently used in our classification schemes. Our results showed that the GAbased fuzzy logic approach is superior to the other two methods. This method enables the classification to achieve a 95.9% of the average recognition rate. Thus the use of GA-based fuzzy logic approach has the potential to become clinically useful for the computer-aided diagnosis of the heart disease.

Keywords

medical images, computer-aided diagnosis, classification

Download

IMEKO-WC-2000-TC13-P347.pdf

DOI

-

IMEKO TC

TC13 - Measurements in Biology and Medicine

Back to the proceedings