Features Extraction Using Difference Operation Method from ECG Signal to Detect Cardiac Arrhythmia

Abstract

Electrocardiogram (ECG) feature extraction plays an important part to find out most of the heart related diseases. An Electrocardiogram wave consist P, QRS and T segments. The electrical impulse generates by Sino-atrial node in the heart. Arrhythmia is one of the common heart related problems which can be detected by the analysis of electrocardiogram of a person. The working of human heart can easily understand with the help of magnitude and intervals value of P, QRS and T waves. The electrocardiogram data used in this research consists of normal and abnormal signals.ECG signals first filtered by IIR notch to remove the artifacts. After filtering, QRS complex of an ECG signal identified. For detection of QRS complex we used DOM (difference operation method). After successfully detection of QRS complex we calculated its R-peak, sharpness, slope and duration. For the classification purpose we used linear classifier in which the ECG data were divided into two partition-one for trained the data called training set in which we used 75% data to trained the classifier and another for test the data called test set in which we used 25% data to test the classifier and classify the normal and arrhythmia signals. The accuracy achieved in this method is 95.30%, sensitivity 96.09% and specificity 96.87%.