ECG Signal Denoising and QRS Complex Detection by Wavelet Transform Based Thresholding

• Main Authors: Swati BANERJEE, Dr. Madhuchhanda MITRA
• Format: Article
• Language: English
• Published: IFSA Publishing, S.L. 2010-08-01
• Series: Sensors & Transducers
• Subjects: DWT; Thresholding; Coefficients; QRS; Approximation; Detail
• Online Access: http://www.sensorsportal.com/HTML/DIGEST/august_2010/P_671.pdf

Biomedical signals like heart waves commonly change their statistical property over time and are highly non stationary signals. For the analysis of this kind of signals wavelet transform is a powerful tool. Electrocardiogram (ECG) is one of the most widely used diagnostic tools for heart disease. Automatic detection of R peaks in a QRS complex is a fundamental requirement for automatic disease identification. This paper presents a novel algorithm and its implementation details for denoising an ECG signal along with accurate detection of R peaks and hence the QRS complex using Discrete Wavelet Transform (DWT) where db6 is selected as the mother wavelet for analysis as it is found to be most similar to the morphology of QRS complexes. Decomposition and selective reconstruction by elimination of higher scale details from the signal, denoises it considerably. Thresholding along with slope inversion method is used for detection of QRS complex. The performance of the system is validated using the 12-lead ECG recordings collected from physionet PTB diagnostic database giving a sensitivity of 99.4 %.