正常小児の各種睡眠段階における脳波活動の自己回帰パワースペクトルと，その構成要素の解析

AUTOREGRESSIVE POWERSPECTRAL AND ITS COMPONENTAL ANALYSIS OF EEG ACTIVITY IN DIFFERENTIAL SLEEP STAGES IN NORMAL CHILDREN

I．はじめに

　1953年，Aserinsky，Kleitmanによつて急速眼球運動を伴う睡眠（レム睡眠・逆説睡眠）が見出されて以来，睡眠には睡眠深度と脳波像とがおおよそ対応するというそれまでの前提は大きくゆるぎ，脳波と睡眠深度との関係が再検討されてきた。そして，睡眠のメカニズムとしては，レム睡眠，ノンレム睡眠そのものを出現させる神経機序と，これら睡眠を始動させ，リズムを形成するより高次の調節機構との両方の面から検討されている。

　ヒトにおける睡眠の神経機序を知る方法は脳波学的研究に負う所が大であるが，視察的脳波所見についてはCaton（1875）^6），Berger（1930）^2）以来幾多の研究がある。

All-night sleep electroencephalograms were re-corded polygraphically in 16 healthy children, from 2 to 15 years old, in order to perform autoregressive powerspectral and its componental analysis in differ-ent sleep stages. From each subject 6 EEG samples of 20 sec length were selected from the various sleep stages for computation of AR?powerspectra and its component, which consisted of several second degree oscillatory activities of alpha, delta, theta, beta rhythms and the first degree mono-scillatory activity of delta wave. The second degree activity was an oscillatory process characterized by frequency of oscillation and damping coefficient (ζ) to indicate decay of amplitude with respect to time. The results were as follows :

1. The relationship between frequency and damping coefficient in various sleep stages : (1) no correlation was found in beta-activity. (2) sigma-activity (activity in the frequency band of sleep spindles) was an obligatory component of sleep stages 2,3 and REM. (3) the damping coefficient of alpha-activity showed a gradual increase as the stage of sleep advanced from stage 1 to stage 4. (4) in contrast to alpha-activity, that of delta-activity decreased as the sleep stage advanced from 1 to 4. (5) the damping coefficient of theta-activity showed the most decreased value in sleep stage 3.

2. The relationship between frequency and mutual information amount in various sleep stages : (1) the information amount in awake and sleep stage REM provided higher value than in other sleep stages. (2) that of delta-activity usually in-creased as the sleep stage advanced from 1 to 4. (3) the ratio of mutual information amount of alpha-activity to that of delta-activity might indi-cate the degree of the depth of sleep stage. Therefore, the depth of the REM sleep was as-sumed to be around that of stage 2.

3. The technique of autoregressive powerspectra and its componental analysis can be usefully applied to the investigation of quantifying EEG activity in sleep.