Deep-ECG : Deep learning framework for ECG signal classification

Abstract

Electrocardiography (ECG) remains the standard method for identifying cardiac irregularities. Automatic detection of abnormalities from the ECG can aid in the early detection, diagnosis, and prevention of cardiovascular diseases. This work presents a novel deep-learning system, named Deep-ECG, designed for the robust detection of Atrial Fibrillation (AF) utilizing three publicly available single-lead ECG databases from PhysioNet (CPSC2018, Georgia, and ECG-Arrhythmia). The proposed system consists of an innovative dual-branch Transformer architecture that accurately classifies atrial fibrillation against normal sinus rhythms. The computational efficiency and high diagnostic sensitivity of the proposed system demonstrate its strong suitability for real-time and outpatient atrial fibrillation screening applications