Feature-Based Image Analysis for Classification of Echocardiographic Images

Masaaki TOMITA

IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences   Vol.E78-A    No.5    pp.589-593
Publication Date: 1995/05/25
Online ISSN: 
Print ISSN: 0916-8508
Type of Manuscript: Special Section LETTER (Special Section of Letters Selected from the 1994 IEICE Fall Conference)
medical image processig,  pattern matching,  computer-aided diagnosis,  echocardiographic images,  

Full Text: PDF>>
Buy this Article

In this letter the classification of echocardiographic images is studied by making use of some texture features, including the angular second moment, the contrast, the correlation, and the entropy which are obtained from a gray-level cooccurrence matrix. Features of these types are used to classify two sets of echocardiographic images-normal and abnormal (cardiomyopathy) hearts. A minimum distance classifier and evaluation indexes are employed to evaluate the performance of these features. Implementation of our algorithm is performed on a PC-386 personal computer and produces about 87% correct classification for the two sets of echocardiographic images. Our preliminary results suggest that this method of feature-based image analysis has potential use for computer-aided diagnosis of heart diseases.