Classification of Congestive Heart Failure Disease, Arrhythmias, And Normal Heart Rhythms Using Deep Learning

Cardiovascular diseases, regardless of their causes, are considered one of the most important causes of death worldwide. For this reason, the issue of time is critical in diagnosing cardiovascular diseases. If a heart disease is diagnosed in a patient by a doctor promptly and then the prescribed solution for treating the disease is properly implemented, not only will it prevent the end of life, but it will also greatly improve Quality of life for people who are facing this issue heart diseases. Given the limitations and challenges in diagnosing cardiac diseases from normal sinus rhythms, the presentation of automated machine learning methods is essential. This article aims to present an efficient method for classifying congestive heart failure disease, arrhythmias, and normal heart rhythms using a combined network of convolutional neural networks and gated recurrent unit networks (CNN+GRU). For this purpose, in the first stage, a dataset consisting of electrocardiogram (ECG) signals in three groups of normal sinus rhythm, congestive heart failure, and cardiac arrhythmias was extracted from the Physio-Net database, and preprocessing and noise removal of the signals was performed. In the second stage, continuous wavelet transform has been used to convert ECG signals into two-dimensional matrices. In the third stage, classification was performed using a combined model of convolutional neural networks and gated recurrent unit networks. Finally, the results obtained from two methods, Convolutional Neural Networks (CNN) and a combined approach of Convolutional Neural Networks and Gated Recurrent Unit (CNN+GRU), have been compared. The results have shown improvement, reduction of errors, and a significant increase in the accuracy of the proposed method in classifying congestive heart failure from other arrhythmias and normal heart rhythms.