Discrimination of PVC based on multiple cardiac cycle fusion and hermite expansion method

Abstract

Signal segmentation plays an important role in Electrocardiogram (ECG) feature extraction. In ECG signals, there are two kinds of dependencies: the dependencies in a single ECG cycle and the dependencies across ECG cycles. The proposed investigation focus on multiple cardiac cycle fusion for ECG feature extraction. Five different feature sets were generated using different ECG segmentation methods and redefinition methods of Premature ventricular contraction (PVC), which were not in medical significance. Hermite coefficients were used as ECG features. The proposed technique was employed to distinguish PVC from normal sinus rhythm (NSR). The data in the analysis were collected from MIT-BIH database. The experimental results show that the features extracted from multiple cardiac cycles classify better than that of single cardiac cycle.