Nakata M, Mukawa J, Fromm G H
University of The Ryukyus, Japan.
Integr Physiol Behav Sci. 1993 Oct-Dec;28(4):343-52. doi: 10.1007/BF02690931.
Since Berger's discovery of the electroencephalogram (EEG), its analysis has been generally restricted to the visual range (upmost 100Hz) and has ignored higher frequency components. One reason should be that there are no reliable methods to distinguish the brain potentials from muscle activity. We have introduced fluctuation analysis, which is popular method especially in the field of basic physiology to clinical electrophysiology. In our previous study, it was declared that power spectral density (PSD) of human high frequency EEG was composed of double Lorentzians and vanished into white level within 1kHz. Then the purpose of this study is to elucidate the "Automated Fluctuation Analysis," which enables us to evaluate these higher frequency components and its physiological meaning especially focused on conscious level from wakefulness to sleep stage 1. Seventy-four scalp recording EEGs in twenty normal subjects were studied. In short, "Automated Fluctuation Analysis" is made of three steps: amplification of EEG signal, A/D conversion and Fast Fourier Transform by signal processor and extraction of Lorentzian parameters. PSD of high frequency EEG was displayed on log-log graph and the algorithm fit to the following Lorentzian formula were mathematically based on Brown & Dennis. S(f) = S1/[1+(f/fc1)2]+S2/[1+(f/fc2)2], where S(f) is PSD (mu V2/Hz) at each frequency (f;Hz), S1 and S2 are the plateau level or zero-frequency power of the initial and second Lorentz, and fc1 and fc2 are the corner or half-power frequency of the initial and second Lorentz, respectively. As results, during wakefulness the PSD of high frequency EEG activity was composed of double Lorentzian fluctuations and the power distribution of S1 value in topographical display was frontal dominant. This pattern of S1 value disappeared and S2 value became lower during sleepiness and the second Lorentz disappeared during sleep.
自从伯杰发现脑电图（EEG）以来，其分析通常局限于视觉范围（最高100Hz），而忽略了更高频率的成分。一个原因应该是没有可靠的方法来区分脑电活动与肌肉活动。我们引入了波动分析，这是一种在从基础生理学到临床电生理学领域都很流行的方法。在我们之前的研究中，宣称人类高频脑电图的功率谱密度（PSD）由双洛伦兹曲线组成，并在1kHz内消失为白色电平。那么本研究的目的是阐明“自动波动分析”，它使我们能够评估这些更高频率的成分及其生理意义，特别是关注从清醒到睡眠1期的意识水平。对20名正常受试者的74份头皮记录脑电图进行了研究。简而言之，“自动波动分析”由三个步骤组成：脑电图信号放大、信号处理器进行A/D转换和快速傅里叶变换以及提取洛伦兹参数。高频脑电图的PSD显示在对数-对数图上，并且基于布朗和丹尼斯的数学算法拟合以下洛伦兹公式。S(f)=S1/[1+(f/fc1)2]+S2/[1+(f/fc2)2]，其中S(f)是每个频率（f；Hz）处的PSD（μV2/Hz），S1和S2分别是初始和第二个洛伦兹曲线的平台水平或零频率功率，fc1和fc2分别是初始和第二个洛伦兹曲线的拐角或半功率频率。结果显示，在清醒期间，高频脑电图活动的PSD由双洛伦兹波动组成，并且在地形图显示中S1值的功率分布以额叶为主。在困倦期间，这种S1值模式消失，S2值降低，并且在睡眠期间第二个洛伦兹曲线消失。
