Heartbeat Classification Using Normalized RR Intervals and Wavelet Features

This study developed an automatic classification system for identifying normal, atrial premature (AP) and premature ventricular contraction (PVC) heartbeats based on normalized RR intervals and wavelet morphological features. The proposed heartbeat classification system consists of signal pre processing, feature extraction, and linear discriminant classification (LDC). First, signal pre processing is applied to remove the high-frequency noise and baseline drift of the original ECG signal. Then the feature extraction includes the normalized RR intervals and the morphological features extracted by the wavelet analysis. Finally, the LDC method is applied to classify the heartbeats according to the extracted features. A total of 48 records obtained from the MIT-BIH arrhythmia database were divided into three datasets for the training and testing of the optimized heartbeat classification system. The testing results show that the normalized RR intervals can enhance the sensitivity for identifying the AP heartbeats in the imbalanced and balanced testing datasets by of 21% and 22%, respectively, and there was an improvement of 18% in the positive prediction accuracy of the normal class in the balanced testing dataset in comparison with non-normalized RR intervals.