Detection of Arrhythmia and Congestive Heart Failure Through Classification of ECG Signals Using Deep Learning Neural Network

Globally, cardiovascular illnesses are the major cause of death. By detecting the electrical activity of the heart as it contracts, electrocardiogram (ECG) data can be used to detect heart abnormalities. Due to individual variability and inherent noise, accurate detection of cardiovascular illnesses is acknowledged to be a difficult task even for human professionals. The current standard of care, visual ECG examination, may lead to individual interpretation and inter-observer predispositions. To overcome the limitations of visual and physical ECG readings, computer-aided diagnosis methods that automatically diagnose ECG have been developed. This study includes modified deep learning neural networks to classify ECG signals into three categories: arrhythmia, congestive heart failure, and normal sinus rhythm. The neural networks used in this work are AlexNet and GoogLeNet. The ECG signals are collected from online sources and are then converted into a scalogram. The scalogram images are divided into train and validation datasets in the ratio of 8:2. The training dataset is then trained using both neural networks. The accuracy of the prediction done by the two architectures are calculated using the confusion matrix. Between the two, GoogLeNet outweighed AlexNet with an accuracy of 96.88%. According to the findings, the GoogLeNet architecture is an effective way for detecting arrhythmias and congestive heart failure.