Hershey, Pennsylvania
New York, New YorkBeijing, China
Biosignal Denoising Techniques

Biosignal Denoising Techniques

Suchetha M. (VIT University, India) and Jagannath M. (VIT University, India)
Copyright: © 2018 |Pages: 12
DOI: 10.4018/978-1-5225-5152-2.ch002
OnDemand:
(Individual Chapters)
Available
$37.50
Current Special Offers
No Current Special Offers
TOTAL SAVINGS: $37.50

Abstract

The main aim of ECG signal enhancement is to separate the required signal components from the unwanted artifacts. Adaptive filter-based ECG enhancement helps in detecting time varying potentials and also helps to track the dynamic variations of the signals. LMS-based adaptive recurrent filter is used to obtain the impulse response of normal QRS complexes. It is also used for arrhythmia detection in ambulatory ECG recordings. Adaptive filters self-modify its frequency response to change the behavior in time. This property of adaptive filter allows it to adapt its response to change in the input signal characteristics. A major problem in adaptive filtering is the computational complexity of adaptive algorithm when the unknown system has a long impulse response and therefore requires a large number of taps. The wavelet transform is a time-scale representation method with a basis function called mother wavelet. In wavelet transform, the input signal is subsequently decomposed into subbands. Wavelet transform thresholding in the subband gives better performance of denoising.
Chapter Preview
Top

1. Introduction To Signal Denoising Techniques

Denoising of biological signal is very seminal to recognize the signal features underlying in noise. Researchers strive to develop an optimum model to eliminate noises of any origin. A corrupted signal containing noise can be estimated by designing a filter that reduces the noise while leaving signals relatively unaffected. Recent techniques are developed for denoising the noisy signals with 50Hz, such filters are introduced by Suchetha (2017). The filter designing in the domain of optimal filtering is pioneered by Wiener (1949) and further modified by Kalman (1960) and Bucy (1961). Better understanding of signal and noise components is required for designing the filters, but adaptive filters can automatically adjust its parameters and no prior knowledge of signal or noise description is needed. Widrow (1975) at Stanford University developed an adaptive noise-cancelling system. Its purpose is to cancel power line frequency interference at ECG amplifier output and recorder output. (Widrow & Stearns 1998).

Traditional adaptive filtering is usually performed in the time domain. Transforms have an important potential for signal processing problems since they can provide a different representation of signals. The Discrete Fourier Transform (DFT) is often used when frequency domain adaptive filtering is applied. However, the DFT is suitable only for stationary signals, for which statistical properties of signals are invariant to a shift of time. Later wavelet transform (WT) has been developed which can be utilised for non-stationary signals analysis. Wavelets come with fast computational algorithms and provide a flexible prototyping environment (Deubechies, 1990; Frazier, 1999). The WT allows extracting features that vary in time, which makes it a useful tool for analysing the signals with transient characteristics. Wavelet shrinkage concepts developed by Donoho and Johnstone (1995) and Bruce and Gao (1996) are the some of the ideas in the field of denoising. Shrinkage analysis analyses the coefficients of empirical wavelet with a threshold and if its magnitude is less than threshold value, then this value is set to zero.

Complete Chapter List

Search this Book:
Reset