A Comparative Deep Learning Approach for Arrhythmia Classification using 1D and 2D ECG Signals Representations

Abstract

Arrhythmias, defined as irregular heartbeats, play a vital role in the early detection of heart disorders using Electrocardiogram (ECG) signals, which capture the heart’s electrical activity and offer valuable insights into cardiac health. In this work, we propose a deep learning approach to compare the performance of one dimensional (1D) raw ECG signals and 2D scalograms, generated through Continuous Wavelet Transform (CWT), using Convolutional Neural Network (CNN) models