Non-linear Analysis of Sleep Electroencephalographic Signals

概要

論文の詳細を見る
Various non-linear dynamical behaviors of brain activity are assessed by means of spectral analysis and chaos analysis on the human electroencephalographic (EEG) signals during sleep and eyes-closed awake conditions. These behaviors are characterized by the slopes of the logarithm of power spectral density plotted as a function of the logarithm of frequency, the three-dimensional phase space trajectories and the correlation dimensions, calculated from EEG time series. It is shown that as the brain becomes more active the slope of the power spectral density decreases and the correlation dimension increases. The correlation dimensions are 9.41±0.20 for eyes-closed awake condition, 6.61±0.81 for light sleep stage, 4.88±0.47 for deep sleep stage and 8.90±0.57 for REM sleep stage, respectively. Since the correlation dimension of EEG is a measure of the complexity of information processing by the brain, our results suggest that complexity might be highest in a relaxed state during alpha activity under eyes-closed condition and during sleep, the complexity decreases with increasing deepness of sleep from light to deep sleep. The relatively high dimensionality of REM sleep may have resulted from the increased information processing of the brain due to dreaming.
東海大学の論文