Perbandingan Skema Dekomposisi Paket Wavelet untuk Pengenalan Sinyal EKG

Abstract

One indicator of a person's health is signal pattern of electrocardiogram (ECG). ECG signals are generated by the heart's electrical activity. ECG signal pattern is recognized by physicians to determine a patient's heart health. Some of the techniques were developed by researchers to automatically recognize the ECG signal. One of the most popular techniques is wavelet transform. In this study, two wavelet packet decomposition schemes for ECG signal recognition are compared to find the best one. The first scheme generates 32 features while the second scheme generates 15 features. Accuracy testing shows that the first scheme produce the best average accuracy of 94.67%, better than the second scheme. Using features selection on the first scheme, four dominant features that produce higher accuracy than using 32 features are obtained. These results indicate that the first scheme is better than the second scheme for ECG signal recognition using wavelet packet decomposition.