Classification of Cardiac Cycles Using a Convolutional Neural Network

One of the main methods for diagnosing cardiac arrhythmias is electrocardiography (ECG). It helps the doctor to understand the causes of various complaints of the patient and to prescribe treatment. There is an urgent need for effective recognition of cardiac pathologies in order to improve the treatment and prevention of cardiac arrhythmias. The doctor is not always able to accurately determine the pathological contractions of the heart due to the fact that the ECG signal is very variable and can be very different in different people with different cardiac disorders. In this paper, proposed an effective algorithm for classification of cardiac pathologies cycles based on images using convolutional neural network. MIT-BIH arrhythmia database was used for training and evaluation of the proposed method. 0.686 – Accuracy, 0.639 – precision, 0.854 – recall and 0.731 – f1-score metric values were achieved.