Neuro-heuristique

SIMPLE AND HIGHER-ORDER ANALYSES OF EEG AND EVOKED POTENTIALS RECORDED IN MAN AND ANIMALS


Higher-order spectral analyses and complex transformations of the analog signals are currently developed for studies of EEG and evoked potentials. In a previous study, visual processing of sinusoidally modulated gratings was investigated in a group of Alzheimer's (AD) patients and a control group. EEG recordings at rest and during visual stimulation were obtained from twenty channels. The power spectrum of the 2nd and 4th harmonic of the stimulation frequency was calculated by FFT at a resolution of 0.25 Hz.

Association of activity between occipital, temporal, parietal, and central regions was measured by intra and inter-hemispheric coherence and phase at the harmonic frequencies. Many functions of the primary visual cortex are preserved in AD patients, whereas occipital to parietal processing and 4th harmonic evoked response are disrupted.

Univariate and multivariate analysis of EEG and evoked potentials is performed by means of second and third order spectra order spectra using conventional and autoregressive estimators.


We are currently developping programs for these analyses under the Matlab programming environment. Power spectrum is able to show the energy distribution among the harmonical components of the signal; cross-spectrum and coherence reveal and quantify the linear relationship between processes recorded on different channels. The time evolution of these analyses illustrates the time-frequency dynamic of the biological signal.







The nonlinear and non-Gaussian characteristics of EEG and, generally, biological signals, suggest the introduction of higher order spectra: the bispectrum detects eventual univariate quadratic nonlinearity, whereas the cross-bispectrum reveals the direction of the propagation of nonlinear phenomena over the different brain regions. Finally the bicoherence is able to quantify objectively the degree of nonlinear inter-channel dependance.