An electrocardiogram classification method based on neural network

Abstract

The ECG is a method for the detection of cardiovascular disease is simple and effective. The ECG. Check that the electricity produced on the heart muscle, cardiac compression. At the point where the heart muscle cells that can create a special type of electricity itself. We call this point that Sinus node electrical current to run through the muscles of the head room on the power that we have called the P wave flow to stop the connection between the atria and ventricles called the AV Node, then electricity will ran down the left and right atria, and the resulting current is called the QRS complex of normal myocardial preview graph. This paper illustrates the classification of electrocardiogram (ECG beats) are proposed trained by feedforward backpropergation method and logistic regression variable selection method. The objective of variable selection is reduce a variable of ECG beat, it will be improving classification, providing faster and avoid over fitting situation. We tested both methods so variable selection method. The ECG Data from MIT-BIH arrhythmia database for classify 5 types. These are atrial premature contraction, Normal, left bundle branch block, right bundle branch block and Premature ventricular contraction. The ECG signal model of cardiac cycle are included P wave, QRS complex, T wave and U wave. A U wave will be invisible by the T wave. So we selected and present the classification and results that make us interested in system design for find new solution for ECG Classification. © 2013 IEEE.