A Quantitative Analysis Approach for Cardiac Arrhythmia Classification Using Higher Order Spectral Techniques

Abstract

Ventricular tachyarrhythmias, in particular ventricular fibrillation (VF), are the primary arrhythmic events in the majority of patients suffering from sudden cardiac death. Attention has focused upon these articular rhythms as it is recognized that prompt therapy can lead to a successful outcome. There has been considerable interest in analysis of the surface electrocardiogram (ECG) in VF centred on attempts to understand the pathophysiological processes occurring in sudden cardiac death, predicting the efficacy of therapy, and guiding the use of alternative or adjunct therapies to improve resuscitation success rates. Atrial fibrillation (AF) and ventricular tachycardia (VT) are other types of tachyarrhythmias that constitute a medical challenge. In this paper, a high order spectral analysis technique is suggested for quantitative analysis and classification of cardiac arrhythmias. The algorithm is based upon bispectral analysis techniques. The bispectrum is estimated using an autoregressive model, and the frequency support of the bispectrum is extracted as a quantitative measure to classify atrial and ventricular tachyarrhythmias. Results show a significant difference in the parameter values for different arrhythmias. Moreover, the bicoherency spectrum shows different bicoherency values for normal and tachycardia patients. In particular, the bicoherency indicates that phase coupling decreases as arrhythmia kicks in. The simplicity of the classification parameter and the obtained specificity and sensitivity of the classification scheme reveal the importance of higher order spectral analysis in the classification of life threatening arrhythmias. Further investigations and modification of the classification scheme could inherently improve the results of this technique and predict the instant of arrhythmia change.

Classification of ECG Signals Using Hybrid Feature Extraction and Classifier with Hybrid ABC-GA Optimization

“…The processing of the information by the heart is reflected in dynamical changes of electrical activity in time, frequency and space. Mostly, features in time [7] and frequency [8] were extracted and combined with efficient classifiers.…”

ECG Signal Denoising and Ischemic Event Feature Extraction using Daubechies Wavelets

“…Most popular Classifier methods employed include linear discriminates, SVM, Gaussian mixture model-based, rule-based rough-set decision system, Bi-spectral analysis technique, supervised classifiers, Hermite functions and self-organizing maps [9].…”

Extraction of Cochlear Non-linearities with the Bispectral Analysis

“…The bispectrum itself has been widely used for the analysis of several biomedical signals [10], such as EEG [11][12] and ECG [13]. We use bispectral analysis to extract nonlinear characteristics from TEOAEs.…”