Robust adaptive techniques for minimization of EOG artefacts from EEG signals

Authors:
S. Puthusserypady;T. Ratnarajah
Affiliations:
Department of Electrical and Computer Engineering, National University of Singapore, Singapore;ECIT, Queen's University of Belfast, Queen's Island, Belfast, UK
Venue:
Signal Processing - Signal processing in UWB communications
Year:
2006

Citing 3
Cited 4

Adaptive filter theory (3rd ed.)

Adaptive filter theory (3rd ed.)
Development of Newton-type adaptive algorithm for minimization of EOG artefacts from noisy EEG signals

Signal Processing
H∞ optimality of the LMS algorithm

IEEE Transactions on Signal Processing

Removing ocular movement artefacts by a joint smoothened subspace estimator

Computational Intelligence and Neuroscience - EEG/MEG Signal Processing
Estimation of Eye Blinking Using Biopotentials Measurements for Computer Animation Applications

ICCVG 2008 Proceedings of the International Conference on Computer Vision and Graphics: Revised Papers
An efficient soft-computing technique for extraction of EEG signal from tainted EEG signal

Applied Soft Computing
Comparative evaluation of existing and new methods for correcting ocular artifacts in electroencephalographic recordings

Signal Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, we propose the application of H∞ techniques for minimization of electrooculogram (EOG) artefacts from corrupted electroencephalographic (EEG) signals. Two adaptive algorithms (time-varying and exponentially-weighted) based on the H∞ principles are proposed. The idea of applying H∞ techniques is motivated by the fact that they are robust to model uncertainties and lack of statistical information with respect to noise [B. Hassibi, A.H. Sayed, T. Kailath, Linear estimation in Krein spaces--part 1: theory & Part II: applications, IEEE Trans. Automat. Control 41 (1996) 18-49]. Studies are performed on simulated as well as real recorded signals. Performance of the proposed techniques are then compared with the well-known least-mean square (LMS) and recursive least-square (RLS) algorithms. Improvements in the output signal-to-noise ratio (SNR) along with the time plots are used as criteria for comparing the performance of the algorithms. It is found that the proposed H∞-based algorithms work slightly better than the RLS algorithm (especially when the input SNR is very low) and always outperform the LMS algorithm in minimizing the EOG artefacts from corrupted EEG signals.