Continuous Wavelet Analysis and Extraction of ECG Features

Abstract

In this chapter, we propose a processing algorithm to measure the features of electrocardiogram records. The continuous wavelet transform is used to analyze the signal in the time-scale domain and evaluate the scalogram for each ECG waves. Thus, we present the literature of wavelet transform and their properties. Next, we propose criteria to choose the mother wavelet more adapted for ECG analysis. Then, the processing algorithm is detailed to extract the features: R-peaks, QRS waves, P waves and T waves. To evaluate the performance, we compute the parameters: sensitivity, predictivity, and error rate for a set of ECG records of the MIT-BIH database. The results of sensitivity Se = 99.84%, positive predictivity P+ = 99.53, and error rate ER = 0.62% demonstrate the effectiveness of the proposed algorithm and its motivation for implementation to improve the processing quality of health systems.