A 5 Class Classification of ECG Using SRCNN and CNN Using MIT-BIH Arrhythmia Dataset

Arrhythmia detection is impossible without computer-based ECG interpretation. The purpose of the study is to present a new deep-learning technique for automatic ECG categorization that can be utilized for arrhythmia identification. Using a dataset of 23 recordings that were randomly selected from a collection of 4000 samples obtained from a heterogeneous group of patients, this work constructed a Super Resolution Convolutional Neural Network to categorize ECG into 5 classes. Based on the nature of the input data and the desired output accuracy, the right classifier is chosen. These Deep Learning results contribute to an ECG classification with strong diagnostic performance comparable to that of cardiologists. The novelty of the paper mainly focuses on the employment of a Super Resolution Neural Network for the predictive classification of arrhythmia. It has been proven that this automated classification of ECG reduces misdiagnosis and provides better accuracy. The tested model provides an accuracy of 87.3%, which is comparatively high for a network just in progressive research.