Classification of ST and Q Type MI variant using thresholding and neighbourhood estimation method after cross wavelet based analysis

TL;DR

This paper introduces a cross wavelet transform based method for ECG analysis that classifies myocardial infarction types with high accuracy by eliminating explicit feature extraction.

Contribution

The proposed technique uses wavelet cross spectrum and coherence for classification, reducing system complexity and improving accuracy over traditional explicit feature methods.

Findings

Achieved 99.43% accuracy in normal vs abnormal classification.

Classified Type I and Type II MI with 88.5% and 87.02% accuracy.

Effectively distinguished MI types using wavelet-based parameters.

Abstract

This paper proposes a cross wavelet transform based method for Electrocardiogram signal analysis where parameters are identified from wavelet cross spectrum and wavelet cross coherence of ECG patterns. Most of the ECG analysing systems use explicit time plane features for cardiac pattern classification. Application of this proposed technique for classification eliminates the need for extraction of various explicit time plane features and hence reduces the complexity of the system. The cross-correlation is the measure of similarity between two waveforms or two time series and the cross examination reveals localized similarities in time and scale. Parameters extracted from Wavelet Cross Spectrum (WCS) and Wavelet Coherence (WCOH) is used for classification. A pathologically varying pattern in QT zone of inferior lead III shows the presence of Inferior Myocardial Infarction (IMI). The Cross Wavelet Transform and Wavelet Coherence is used for the cross examination of single normal and abnormal (IMI) beats. A normal template beat is selected as the absolute normal pattern. Computation of the WCS and WCOH of the selected normal template and various other normal and abnormal beats reveals the existence of variation among patterns under study. The Wavelet cross spectrum and Wavelet coherence of various ECG patterns shows distinguishing characteristics over two specific regions R1 and R2, where R1 is the QRS complex location and R2 is the T wave region. Parameters are identified for classification of Type 1 IMI (non Q type, with ST elevation and attenuated QRS complex) and Type 2 IMI (Q type MI with deep Q and inverted T) and normal subjects. Accuracy of the proposed classification method is obtained as 99.43% for normal and abnormal class and 88.5% and 87.02% for Type I and Type II respectively.