A Review on Analysis of Cardiac Arrhythmia from Heart Beat Classification

Abstract

Cardiac arrhythmias are abnormal heartbeats that might be irregular, excessively rapid, or too slow. Cardiac arrhythmia occurs when electrical impulses in the heart malfunction. The early diagnosis of cardiac arrhythmias may aid in the cure or treatment of the illness regarding heart. This situation can be detected through biomedical signals such as Electrocardiograms (ECGs), These signals are monitored to identify heart electrical activity through sensors implanted over human body's chest. Heart problems will be recognized by analyzing these ECG signals. The anomaly will be categorized based on the kind of ECG found. This has been considered by numerous researchers, who have suggested some techniques for cardiac arrhythmias detection. The strategies for cardiac arrhythmias detection have been published in the last four years (2018-2022), are the primary topic of this paper. In recent years Machine Learning (ML) and Deep Learning (DL) made remarkable accomplishments in signal processing field. While comparing with traditional neural networks (NN), deep neural networks (DNN) are capable of automatic feature extraction, and recognize intricate patterns of data and also eliminate complex signal processing. Few DNN include Support Vector Machine (SVM), recurrence network, Deep Belief Network (DBN), Long Short-Term Memory (LSTM), Convolutional Neural Network (CNN), Gated Recurrent Unit (GRU). Based on the study, it is observed that CNN is mostly referred as the appropriate feature extraction technique. While using CNN, GRU and LSTM separately, DL techniques demonstrated great accuracy in correctly classifying the major Cardiac Arrhythmia "Atrial-Fibrillation (AF), SupraVentricular Ectopic Beats (SVEB), Ventricular Ectopic Beats (VEB)".