Dynamic time warping as a novel tool in pattern recognition of ECG changes in heart rhythm disturbances

We present a novel method for the classification and identification of electrocardiograms (ECGs) of various heart rhythm disturbances. This is an essential step in the automatic analysis of heart rhythm disturbances. Dynamic time warping (DTW) is used for this purpose. DTW is utilized successfully in speech recognition. Wavelet analysis is used in some implantable cardioverter defibrillators currently for ECG waveform recognition and classification purpose. The simulations of time-series ECG data of various rhythm disturbances are produced. Normal sinus rhythm ECG templates are compared to the simulated rhythms by both methods. DTW analysis successfully differentiates the ECGs of various arrhythmias. Of note, DTW is able to differentiate ventricular tachycardia from supraventricular tachycardia unlike wavelet analysis. Differentiation of these two rhythm types has significant clinical implications. DTW can potentially be used for automatic pattern recognition of ECG changes representative of various rhythm disturbances.