Classification of Myocardial Infarction with ECG Signals using RESNET

Abstract

One of the most dangerous cardiovascular diseases is Myocardial Infarction for human beings. It
is commonly recognized as heart attack. Myocardial Infarction (MI) is caused by the impairment of
cardiac muscle. The Electro Cardio Gram logs the electric pulse of the heart and these data are able
to reflect the aberrant actions of the heart. When there is a block in veins of heart the flow of oxygen
to cardiovascular muscle gets stopped which make hurt to heart muscles. The irrevocable harm may
prompt to mortality. With the quick upgradation of wearable devices and favorable
electrocardiogram (ECG) clinical apparatus, it is reasonable and feasible to identify and screen
dead tissue in heart muscles. Fast and accurate identification of Myocardial Infarction is necessary
to decrease the mortality rate. Physical Electro Cardio Gram understanding needs skilled and is
leads to inter-observer variability and more time is consumed for it. The physical interpretation may
change from one individual to another. Automatic detection of MI on ECG can be done with a
computer aided diagnosis. The aim is to give an algorithm for the ML detection that process directly
on information of ECG. The images are classified using a deep learning model with ResNet and
images are identified to check whether there is a presence of Myocardial Infarction or not. The
ResNet is capable of calculating an image accurately. The proposed ResNet model yields 96%
accuracy for the given dataset.