Prediction of ventricular tachycardia using morphological features of ECG signal

Abstract

Ventricular tachyarrhythmia particularly ventricular tachycardia (VT) and ventricular fibrillation (VF) are the main causes of sudden cardiac death in the world. A reliable predictor of an imminent episode of ventricular tachycardia that could be incorporated in an implantable defibrillator capable of preventive therapy would have important clinical utilities. As variability of T wave, ST segment and QT interval are indicators of cardiac instability, these changes can lead us to develop accurate predictor for VT. In this study, we present an algorithm that predicts VT using morphological features of electrical signal of ventricles activity obtained from Electrocardiogram (ECG). Changes in T wave, ST segment, QT interval and numbers of premature ventricular complexes(PVCs) are considered as effective indicators of VT. Classification of selected features by a Support Vector Machine (SVM) can identify hidden patterns in ECG signals before VT occurrence. Evaluation of this algorithm on 40 recods of VT patient and 40 control records shows that the proposed algorithm can reach sensitivity of 88% and specificity of 100% in VT prediction.