Combining continuous wavelet transform and Teager-Kaiser Energy operator for ECG arrhythmia detection

Computer-aided diagnosis of Electrocardiographic (ECG) signal consists of processing the signal to extract parameters in time, frequency, scale, etc. Then distinct parameters that provide significant differences between normal and arrhythmic ECG beats are selected as a feature vector. This feature vector is fed to either a supervised or an unsupervised classifier for detection. Seeking significant features is an important step for the success of the classifier. In this paper, we address combining both continuous wavelets transform (CWT) and the Teager-Kaiser Energy (TKE) for feature extraction. This is for developing a method for the detection of premature ventricular contraction (PVC) beats in ECG. The proposed method has been tested by the MIT-BIH arrhythmia database. The method has shown promising detection accuracy of 100% in classifying normal and PVC beats.