Automated Detection of Myocardial Infarction with Scalogram Technique and Deep Convolutional Neural Network

Electrocardiogram (ECG) is a transthoracic interpretation of the heart’s electrical activity over time. It is used to diagnose myocardial infarction (MI), coronary artery disorders, and other heart disorders. MI results from coronary artery blockage due to the death of cardiac muscle tissue. The early stage of diagnosis and prompt treatment can save lives. This paper presents a MI detection model based on deep learning method using scalogram technique. The proposed method uses a scalogram to transform 1D ECG data into 2D scalogram images. Last layer of the proposed model uses pretrained DenseNet to classify six types of MI with 98.63% accuracy. The model developed in this paper is less complex because it doesn’t need additional noise filtering and handcrafted feature extraction steps. According to the study, a portable device might be developed using chest that leads to serve as an automated diagnosis tool.