Abstract—Multiscale Principal Component Analysis (MSPCA) is applied for quality controlled denoising of Multichannel Electrocardiogram (MECG) signals. Wavelet transform of MECG signals disseminates clinical information content into different wavelet subbands or scales. Collecting wavelet coefficients of all ECG channels at a wavelet scale multivariate data matrices are formed. Principal Component Analysis (PCA) is performed on these matrices for signal denoising. The desired quality of processed signals is achieved by selecting the principal components (PC) based on energy features in selected wavelet subband matrices. To control the quality of denoised signals, the number of PC selection is based on cumulative percentage of total variation of variances. The choice of multiscale matrices and selection of eigenvalues preserve the desired energy in the processed signals. Quantitative performance is measured using input and output Signal-to-Noise Ratio (SNR). Signal distortion metrics are evaluated using Percentage Root Mean Square Difference (PRD) and Wavelet Energy based Diagnostic Distortion (WEDD) measures. SNR improvement of 31.12 dB has been found with better denoising effect using database of CSE Multilead Measurement Library.
Index Terms—Denoising, ECG, MSPCA, PCA, PRD, WEDD.
The authors are with the Department of Electronics and Electrical Engineering, Indian Institute of Technology Guwahati, India (e-mail: lns@iitg.ernet.in.; e-mail: samaren@iitg.ernet.in; E-mail: anilm@iitg.ernet.in.)
Cite: L. N. Sharma, S. Danadapat, and A. Mahanta, "Multiscale PCA based Quality Controlled Denoising of Multichannel ECG Signals," International Journal of Information and Electronics Engineering vol. 2, no. 2, pp. 107-111, 2012.