Arrhythmia Classification on Electrocardiogram Signal Using Convolution Neural Network Based on Frequency Spectrum

Heart disease is the leading cause of death in the world. To find out heart disease early, it can be detected by examining the presence or absence of arrhythmias. Arrhythmia is an abnormal heart beat rhythm, can beat too fast, too slow, or beat with irregular patterns, so that the arrhythmia has many types. To diagnose arrhythmias, one method that can be used is by analyzing ECG (Electrocardiogram) signals. Currently, doctors and medical personnel analyze ECG signals manually. Because the number of cardiologist paramedics is far less than the number of patients, patients need hardware or software to analyze the heart independently. With the development of technology in this era, there is a technology called Deep Learning. Deep Learning is a development of Machine Learning. In this paper, we proposed one method of Deep Learning, namely Convolutional Neural Network (CNN), is used to classify 5 types of arrhythmias on ECG signals, that are: Normal Beat (NOR), Right Bundle Branch Block (RBBB), Left Bundle Branch Block (LBBB), Premature Ventricular Contraction (PVC) and Fusion of Ventricular and Normal (FVN). Evaluating the performance of our CNN architecture, we tested it to classify the heart beat in the MIT BIH Arrhythmia database. The performance results of our proposed have highest accuracy is 98.6% with the following details: 713 Normal Beat samples can be detected correctly (100%), RBBB 689 samples (96.63%), LBBB 710 samples (99.58%), FVN 713 samples (100%), and PVC 690 samples (96.77%).