Localization and Classification of Myocardial Infarction Based on Artificial Neural Network

In this article, has been developed Convolutional Neural Network (CNN) architecture for classifying and determining the localization of myocardial infarction based on electrocardiogram (ECG). Training process is carried out on deep learning toolbox of MatLab. For the training, the database based on a 12-lead ECG device is developed. Database consisted of 11 classes: 10 classes belong to patients of myocardial infarction and one class's data taken from healthy persons. All data of total classes saved on. mat file form, and each class consisted of different amount of .mat files. For each class, network accuracy calculated separately and average result determined. The training result achieved at 98.47%. The results also compared to the obtained results of other researchers. The specifics of each taken results are also discussed.