tailieunhanh - Biomedical Engineering 2012 Part 12

Tham khảo tài liệu 'biomedical engineering 2012 part 12', kỹ thuật - công nghệ, cơ khí - chế tạo máy phục vụ nhu cầu học tập, nghiên cứu và làm việc hiệu quả | 432 Biomedical Engineering 24 Nonlinear Projective Filtering of ECG Signals Marian Kotas Silesian University of Technology Institute of Electronics Poland 1. Introduction Since the beginnings of the noninvasive electrocardiology development of the methods for reconstruction of the electrocardiogram ECG embedded in noise has stimulated the progress in the field. The predominant types of noise are baseline wander powerline interference electromyographic EMG noise and motion artifacts. Among them EMG and motion artifacts are the most difficult to be suppressed. It stems from the fact that their frequency spectra overlap that of the desired ECG which makes the classical band-pass filtering ineffective. To mitigate the problem the synchronized averaging was introduced. The method assumes that the ECG signal is repeatable and that the noise is additive independent and of zero mean. Linear time-alignment Jane et al. 1991 of the respective ECG beats and construction of an average one results in significant suppression of the noise however the inter-beat variability of the cardiac cycles is lost. In many applications it is advantageous to suppress noise while preserving the variability of the desired signal morphology. In such cases it is more advantageous to create a space of possible shapes instead of constructing an average template. This task can be accomplished with the help of principal component analysis PCA Olmos et al. 1999 Kotas 2006 . In Paul 2000 the signal subspaces were constructed in the discrete cosine transform domain. However modeling ECG beats with the help of the intrinsically linear PCA allows a limited success only Kotas 2006 . In Hu Nenov 2006 construction of the signal subspaces was preceded with the nonlinear alignment of ECG beats. New capabilities emerged when the method of nonlinear state-space projections NSSP was applied to ECG processing Schreiber Kaplan 1996 . Although this method also performs locally linear projections globally the .