Recognition of Myocardial Ischemia Electrocardiogram Signal Based on Deep Learning

In this study, a 28-layer deep residual network model is designed to recognize myocardial ischemic Electrocardiogram (ECG) signals, and has been verified on a data set containing 10,000 clinical 8-lead ECG signals. In addition, in order to compare the effects of recognizing myocardial ischemic ECG signals between deep residual network and waveform-based feature recognition algorithm, the waveform-based feature recognition algorithm is used to perform myocardial ischemic binary classification experiment on the same data set. The experimental results show that the deep residual network algorithm can achieve 99.75% accuracy when recognizing myocardial ischemic ECG signals and normal ECG signals, and the accuracy rate reaches 95.2% when recognizing myocardial ischemic ECG signals and diseased ECG signals. The deep residual network algorithm is superior to the waveform-based feature recognition algorithm in classification accuracy, sensitivity and specificity.