Application of multiscale principal component analysis (MSPCA) for enhancement of ECG signals

The electrocardiogram (ECG) is an electrical manifestation of contractile activity of the heart. The presence of parasite interference signals, like power line interference (PLI), Electromyogram (EMG) or myopotential signals and baseline drift interferences, could cause serious problems in the registration of ECG signals. Many a time, they pose problem in modern control and signal processing applications by being a narrow in-band interference near the frequencies carrying crucial information. The wavelet transform is a powerful time-frequency analysis tool analysis of complex non stationary signals. Principle Component Analysis (PCA) is a standard tool in modern data analysis because it is simple non-parametric method for extracting relevant information from complex data sets. In multi scale Principal Component Analysis (MSPCA), the PCA's ability to decorrelate the variables by extracting a linear relationship and wavelet analysis are utilised. This paper presents MSPCA method proposed for the enhancement of ECG signals, so that the enhanced signal can be used for clear identification of arrhythmias. In MSPCA, the principal components of the wavelet coefficients of the ECG data at each scale are computed forst and are then combined at relevant scales. In this application for ECG enhancement, the MSPCA served as a powerful tool when addressing problems related to noise reduction. Results revealed that Daubenchies based MSPCA out performed the basic wavelet based processing for ECG signal enhancement.