Dumermuth G
Schweiz Arch Neurol Neurochir Psychiatr. 1974;115(2):175-92.
Quantification and analysis are of major interest in modern electroencephalography. During recent years developments in mathematical-statistical analysis of time series and in electronics and computer technology have brought significant advances. A common feature of almost all procedures proposed so far is that they treat the EEG as a random variable. Various statistical descriptors are estimated either in the time or in the frequency domain, for one or several simultaneous EEG channels. This paper reviews the various methods available today to quantize and analyze the main categories of EEG activities, i.e. (1) spontaneous non-paroxysmal activities, (2) spontaneous paroxysmal activities and (3) activities evoked by external sensory stimulation. Due to personal preferences of the author, emphasis is put on analysis methods in the frequency domain. The main problem today lies much less in the technical installation of the one or the other analytical method than in the treatment of artifacts and in postprocessing the large amount of analyzer output. Therefore, applications must be restricted to carefully selected special questions. Clinical electroencephalography is still far from automatization.
量化与分析是现代脑电图学的主要研究内容。近年来，时间序列的数理统计分析以及电子和计算机技术的发展取得了显著进展。迄今为止提出的几乎所有方法的一个共同特点是，它们将脑电图视为一个随机变量。针对一个或几个同步的脑电图通道，在时域或频域中估计各种统计描述符。本文回顾了目前用于量化和分析脑电图活动主要类别的各种方法，即(1) 自发性非阵发性活动，(2) 自发性阵发性活动，以及(3) 外部感觉刺激诱发的活动。由于作者的个人偏好，重点放在频域分析方法上。如今，主要问题与其说是某种分析方法的技术安装，不如说是伪迹的处理以及对大量分析仪输出的后处理。因此，应用必须限于精心挑选的特殊问题。临床脑电图学离自动化仍有很大距离。
