ACCURATE DETECTION AND CASSIFICATION OF VENTRICULAR ABNORMAMLITIESS USING MORPHOLOGICAL FEATURES

Abstract

Electrocardiogram (ECG) is a bioelectrical signal which presents the electrical activity of different parts of heart in time domain. The accurate study of this signal leads to precise detection of ventricular arrhythmias that are serious problems in today's world and might lead to Sudden Cardiac Death (SCD). This research attempts to achieve accurate and fast detection and classification of ventricular abnormalities from morphological features using cross correlation between normal sinus beat and abnormal beats. Cross correlation between RR intervals and TT intervals in zero lag was used because of separation between normal sinus rhythm and T wave alternans. Specific morphological features were attained for T wave alternans and normal sinus rhythm signals. The classification accuracy of normal sinus rhythm with ventricular tachycardia, ventricular flutter, ventricular fibrillation, premature ventricular contraction and escape beat rhythm were obtained as 100%, 92.8%, 86.8%, 90.81%, 96.28% and 91.8%, respectively. Classification accuracy of T wave alternans was achieved as 95.03%.