Effect of independent component analysis on multifractality of EEG during visual-motor task

Authors:
D. Popivanov;S. Jivkova;V. Stomonyakov;G. Nicolova
Affiliations:
Motor Control Laboratory, Institute of Physiology, Bulgarian Academy of Sciences, Sofia, Bulgaria and Laboratory of Neuroscience, New Bulgarian University, Sofia, Bulgaria;Motor Control Laboratory, Institute of Physiology, Bulgarian Academy of Sciences, Sofia, Bulgaria;Motor Control Laboratory, Institute of Physiology, Bulgarian Academy of Sciences, Sofia, Bulgaria;Motor Control Laboratory, Institute of Physiology, Bulgarian Academy of Sciences, Sofia, Bulgaria
Venue:
Signal Processing - Neuronal coordination in the brain: A signal processing perspective
Year:
2005

Citing 4
Cited 2

Characterization of Signals from Multiscale Edges

IEEE Transactions on Pattern Analysis and Machine Intelligence
Independent component analysis, a new concept?

Signal Processing - Special issue on higher order statistics
An information-maximization approach to blind separation and blind deconvolution

Neural Computation
Superefficiency in blind source separation

IEEE Transactions on Signal Processing

Multifractal correlation characteristic for radar detecting low-observable target in sea clutter

Signal Processing
Multifractal analysis of intracranial EEG in epilepticus rats

ICONIP'11 Proceedings of the 18th international conference on Neural Information Processing - Volume Part I

Quantified Score

Hi-index	0.00

Visualization

Abstract

In visual-motor task, blinks, muscle and other short-term artifacts may corrupt the correct determination of the fractal properties of the brain signals that reflect task performance. We tested whether independent component analysis (ICA) is a necessary preprocessing tool to determine fractal properties of amplitudes of decomposed EEG records in theta, alpha, beta and gamma oscillations. The subjects were required to track a moving spot on a city map displayed on a computer screen by pushing forward a joystick in two experimental conditions: the spot moved in steps either with constant or fractal time intervals. The wavelet transform modulus maxima (WTMM) method was applied to estimate the local fractal dimensions D(h) for recovered EEG via reduced independent components (ICs) localized inside and outside the brain. The measure of fractality, i.e. the Hoelder exponent hDmax, was statistically estimated among experimental conditions. We established multifractality for extracted IC per se, specific for filtered oscillations in both experimental conditions: long-range correlation for theta and alpha, and anticorrelation for beta and gamma. Similar results were obtained for filtered versions of recovered EEG. Multifractal scaling, specific for lower and higher EEG oscillations, proved to be very stable intrinsic feature for the activity of large brain areas. The external events (task conditions) and the extended number of ICs, including those at the boundary line of the brain, did not have influence upon the scaling, although their effects might be statistically different for a given filtered oscillation.