The Studies Of Sleep Staging And Scaling Behavior Of EEG In Rats

As one of the most complex systems in nature, brain always exhibits rich spatiotemporal dynamic characteristics. Electroencephalogram (EEG) oscillations with different frequencies reflect changes of these properties, and various vigilance states in sleep-wake cycles represent different functional states of brain. Thus to explore scaling behaviors and their underlying mechanisms of various EEG oscillations during different sleep stages will help to further understand dynamic characteristics of brain.Based on animal experiments, this thesis firstly optimizes a non-threshold based algorithm using a single EEG channel. This algorithm is a useful strategy for sleep staging in rats for long-term study of sleep. Then scaling behaviors of EEG oscillations from different derivations, differences of scaling exponents among various EEG oscillations during different vigilance states and the oscillation which plays a key role during sleep transitions, are estimated by detrended fluctuation analysis (DFA). And finally, the underlying mechanisms of scaling behaviors are investigated by the Markov process amplitude (MPA) EEG model. The following studies are presented and discussed:1.The accuracies of sleep staging at different electrode locations is studied by an algorithm using a single EEG channel, and the site of the frontal electrode in a frontoparietal bipolar electrode located at the anterior midline is proved the optimum.2.Comparative studies are carried out across two different non-threshold based algorithms using EEG only and two different sets of coordinates of electrodes. The results indicate that the accuracy of sleep staging, especially for rapid eye movement (REM) sleep, could be improved by increasing classification parameters, optimizing the filter bandpass and the coordinates of an electrode pair for a single EEG channel.3.The effects of brain music on sleep and arousal are examined. The results demonstrate that exposure to the brain music increases arousal levels and decreases sleep in rats, and that the underlying mechanisms of decreased non-rapid eye movement (NREM) might be different from that of REM sleep.4.The properties of various EEG oscillations are studied by comparing the spatial differences of the long-range temporal correlations (LRTC) after calculating their scaling exponents using DFA. The results indicate that the scaling exponents of an oscillation acquired from different derivations differ significantly in each vigilance state; and these differences might be related to sleep stage-dependent generators of various neuronal processes.5.The scaling exponents and LRTC of various EEG oscillations during different vigilance states and the underlying mechanisms of scaling behaviors are discussed using DFA and MPA EEG model respectively. The results indicate that: (1) During a given vigilance state, scaling exponents increase with rising frequency of EEG oscillations. The changes of scaling exponents imply that distinct neural networks and mechanisms might underlay different EEG oscillations; and for each EEG oscillation, scaling exponents among different vigilance states are also significantly different from each other. These differences might be related to internal and external events accompanying sleep-waking cycles. Furthermore, the results of scaling exponents or LRTC reveals that a given EEG oscillation might be caused by the concerted action of regularity and randomicity of its underlying neuronal process, thus the changes of scaling exponents might indicate the transitions of functional states of brain. (2) Low-frequency oscillations and high-frequency oscillations might adopt different dynamical mechanics, i.e. 1/ f for the former and 1/ f 2 for the latter. It is supposed that low-frequency oscillations might play a key role in establishing and maintaining the most fundamental functions and states of brain, while high-frequency oscillations might be mainly involved in relatively higher-level functions. These functional assignments allow brain to operate reliably and simultaneously at multiple spatiotemporal scales. (3) Because the power spectrums of low frequency oscillations with LRTC are dominant in brain, it is reasonable to infer that 1) brain might be a self-organized criticality (SOC) system, 2) sleep alternations might be achieved by means of the processes similar to avalanches, and 3) low-frequency EEG oscillations might play a key role in sleep transitions.