Pattern based ECG diagnosis

Abstract

A system with a complex data structure which provides for diagnosis of multiple disorders of the heart is introduced. In the proposed data structure, a tree is used to represent a waveform. The normal and abnormal electrocardiogram (ECG) waveforms are represented by a tree structure. The degree of mismatch between the two trees in terms of node splitting and merging operations are calculated. The three distances which are a measure of the structural deformation of the waveforms form the background medical information required for diagnosis. A key characteristic of the system is the congenial man-machine interface. The system is capable of explaining to the user the reasoning process for its diagnosis. The model is independent, thereby allowing expansions to include disorders from other disciplines.<>