Cardiac Arrhythmia Classification using Deep Convolutional Neural Network and Fuzzy Inference System

Arrhythmia is a cardiac disorder due to abnormal heart rhythm. Cardiologists used a technique known as an electrocardiogram (ECG) to diagnose arrhythmias or heart disorders. Noise, non-stationary ECG signals, and irregular heartbeat make arrhythmias difficult to diagnose. Therefore, a computer-assisted diagnostic system with improved detection accuracy is required. The accurate identification of arrhythmic types may efficiently stop cardiovascular disease and minimize the risk of death. In recent decades, Neural networks (NNs) and fuzzy systems have shown to be effective in intelligent systems. In this paper, a Fuzzy Inference System (FIS)-based Three-dimensional neural network (3D-CNN) for binary classification (normal and arrhythmia) of 2-lead ECG signals are developed and tested in the MIT-BIH dataset. To accurately illustrate the classifying results, common metrics such as F-score, recall, precision, specificity, sensitivity, and accuracy. The proposed approach is tested using MATLAB and the results are evaluated.