Wavelet-based Techniques Applied to Digital Processing of ECG Signals

Electrocardiogram (ECG) is an important biomedical signal for diagnosing heart diseases, but now it can have different other applications, such as using it as a stress recognition biomarker. Taking into account that the ECG signal is always overlapped with noise generated by muscles, body movement, electrodes skin contact, breathing and electronics, a de-noising stage must be added. This research paper analyzes and compares the noise removal by using different decomposition levels of undecimated wavelet transform (UWT) and discrete wavelet transform (DWT) based on different types of mother wavelets like orthogonal (Haar, Daubechies, Coiflets, Symmlet) and biorthogonal, whereas the other artifacts that are low frequency carriers have been already removed from the signal. During the analysis phase for each of the de-noising methods, several parameters are modified to check the accuracy and performance in a more elaborated way. The raw ECG data used in the study has been obtained by using a proposed data acquisition system.