Using bispectral analysis in OSAS estimation

In this study the possibility of estimation of Obstractive Sleep Apnea Syndrome (OSAS) through electroensephalographic (EEG) signals using higher order spectra was investigated. Biological structures usually exhibit nonlinear and non-Gaussian distributed characteristics which consequently generate signals embracing nonlinear components as well as phase relations occurring between componets oscillating in different frequencies whose phase couples together over a limited period of time, so called Quadratic Phase Coupling (QPC). In this case the second order power spectrum may not reflect the real characteristics of the biological system. Therefore, the bispectrum analysis was achieved to characterize the nonlinearities and QPCs in OSAS and normal EEGs. Through this analysis the differences in OSAS EEG and normal EEG were comparatively uncovered. From the analysis it was understood that the bispectrum can be used for estimation of OSAS from patients´ EEG signals.