Uchida S, Matsuura M, Ogata S, Yamamoto T, Aikawa N
Department of Psychophysiology, Tokyo Institute of Psychiatry, Japan.
J Neurosci Methods. 1996 Jan;64(1):1-12. doi: 10.1016/0165-0270(95)00115-8.
This article critically reviews 8 computer implementations of Fujimori's method for EEG waveform recognition, with methodological considerations for the application of this method to the analysis of all-night sleep EEG. Fujimori's method has been considered one of the most appropriate waveform analyses for EEG. This kind of analysis is advantageous for measuring frequency and amplitude of each EEG wave separately. However, current implementations have drawbacks which must be resolved before they can be used on all-night sleep EEG. An optimal sampling rate should be determined which is appropriate to the purpose of analysis. Amplitude thresholds for wave recognition, which are now set arbitrarily, should also be improved. Measurement of waves in higher orders of superimposition is also necessary, although existing systems are limited to the second order. Additional algorithms, such as for the separate detection of sleep slow waves, may be useful. Further applications for Fujimori's method are suggested.
本文批判性地回顾了用于脑电图（EEG）波形识别的藤森方法的8种计算机实现方式，并探讨了将该方法应用于整夜睡眠脑电图分析时的方法学考量。藤森方法被认为是脑电图最适用的波形分析方法之一。这种分析方法有利于分别测量每个脑电波的频率和振幅。然而，目前的实现方式存在缺陷，在用于整夜睡眠脑电图之前必须加以解决。应确定适合分析目的的最佳采样率。目前随意设定的波形识别幅度阈值也应改进。尽管现有系统仅限于二阶叠加，但对高阶叠加波形的测量也是必要的。诸如用于单独检测睡眠慢波的附加算法可能会有用。文中还提出了藤森方法的进一步应用。
