Frequency-domain features for ECG beat discrimination using grey relational analysis-based classifier

This paper proposes a method for electrocardiogram (ECG) heartbeat discrimination using novel grey relational analysis (GRA). A typical ECG signal consists of the P-wave, QRS complexes and T-wave. We convert each QRS complexes to a Fourier spectrum from ECG signals, the spectrum varies with the rhythm origin and conduction path. The variations of power spectrum are observed in the range of 0–20 Hz in the frequency domain. To quantify the frequency components among the various ECG beats, GRA is performed to classify the cardiac arrhythmias. According to the AAMI (Association for the Advancement of Medical Instrumentation) recommended standard, heartbeat classes are recommended including the normal beat, supraventricular ectopic beat, bundle branch ectopic beat, ventricular ectopic beat, fusion beat and unknown beat. The method was tested on MIT–BIH (Massachusetts Institute of Technology–Beth Israel Hospital) arrhythmia database. Compared with other artificial intelligence (AI) methods, the results demonstrate the efficiency of the proposed noninvasive method, and also show high accuracy for detecting ECG signals.