Classification of Fibrillation Subtypes with Single-Channel Surface Electrocardiogram

Abstract

Atrial fibrillation (AF) and ventricular fibrillation (VF) are complex heart rhythm disorders with increasing prevalence. Mechanisms sustaining these arrhythmias are different, and subsequently, the required treatments options differ. Although many algorithms have been developed for differentiating fibrillation from normal sinus rhythm, very few methods exist to differentiate between different forms of AF and VF from surface electrocardiogram (ECG). To address the issue, we propose a novel ECG classification method to differentiate fibrillation that is completely chaotic from forms where it is organized with key driving sites. Differentiating fibrillation organisation from ECGs may aid patient selection, and identify those who may benefit from targeted ablation treatment. Evaluation using real-world data sets based on rat VF model shows that the proposed method could recognise the correct Fibrillation subtype from the single-channel electrocardiogram with an accuracy of 88.89$\%$.