Cardiac arrhythmia classification in 12-lead ECG using synthetic atrial activity signal

Abstract

Analysis of the ECG signal is the prevalent method for diagnosing cardiac arrhythmia. In order to achieve a precise diagnosis, the physician must carefully examine the quantity, location, and relations between the ECG signal elements, with emphasis given to the atrial electrical activity (AEA) wave characteristics. Nevertheless, in some cases the AEA-waves are hidden in other waves, and in order to classify the correct arrhythmia an invasive procedure is performed. We propose a fully automated computer-based method for arrhythmia classification, based on our recently developed AEA detection algorithm, combined with two extracted rhythm-based features and a clinically oriented set of rules. Twenty-nine patients presenting atrioventricular nodal reentry tachycardia, atrioventricular reentry tachycardia, sinus tachycardia, atrial flutter, and sinus rhythm were studied. The arrhythmia classifier achieved 92.2% accuracy, 83.9% sensitivity, and 94.9% specificity.