Author:
Talbi Mourad
Affiliation:
Research and Technology Center of Energy of Borj Cedria, Tunisia
Keyword(s):
Electrocardiogram, R Peaks, Multi-Scale Product, Undecimated Wavelet Transform, Principal Component Analysis.
Related
Ontology
Subjects/Areas/Topics:
Applications and Services
;
Biomedical Engineering
;
Biomedical Signal Processing
;
Computer Vision, Visualization and Computer Graphics
;
Detection and Identification
;
Medical Image Detection, Acquisition, Analysis and Processing
;
Real-Time Systems
;
Wavelet Transform
Abstract:
In this paper, we propose a new Electrocardiogram (ECG) Denoising technique based on Principal Component Analysis (PCA) and using R peaks detection. This technique consists at first step in cutting the entire ECG signal into frames then the denoising is performed frame by frame by using PCA. Each frame is located between two successive R peaks. The R peaks detection is performed by using a new detection method based on multi-scale product of the undecimated wavelet coefficients. The Reconstructed ECG signal is obtained by concatenating all the denoised frames. The evaluation of the proposed technique is performed by comparing it to the denoising technique based on PCA and applied to the entire noisy ECG signal. The two techniques are tested on four ECG signals taken from MIT-BIH database. The used criteria in this evaluation of these two techniques are the SNR improvement and the mean square error (MSE). The obtained results from this evaluation show clearly that the denoising techni
que based on PCA and applied to the entire noisy ECG signal, is slightly better than the proposed technique. However this latter has the advantage of working in real-time because the processing is performed frame by frame and not on the entire noisy ECG signal. Concerning the new proposed technique of R peaks detection, it is very accurate because it permits a perfect reconstruction of the ECG signal when concatenating all the frames.
(More)