Feature Extraction Based on Circular Summary Statistics in ECG Signal Classification

Abstract

In order to explore new patterns for classification of cardiac signals, taken from the electrocardiogram (ECG), the circular statistic approach is introduced. Features are extracted from instantaneous phase of ECG signal using the analytic signal model based on the Hilbert transform theory. Feature vectors are used as patterns to distinguish among different ECG signals. Five types of ECG signals are obtained from MIT-BIH database. Preliminar results shown that the proposed features can be used on ECG signal classification problem.