O'Neill W D, Oroujeh A M, Merritt S L
Department of Electrical Engineering and Computer Science, University of Illinois at Chicago 60607-7053, USA.
IEEE Trans Biomed Eng. 1998 Mar;45(3):314-22. doi: 10.1109/10.661156.
The pupil light reflex has a long history of being able to indicate states of mental arousal, ranging from sleepiness to concentrated cognitive effort. Such mental states have usually been inferred from pupil diameter or pupil area movements relative to some reference; sleepiness, for example, is characterized by a smaller than average pupil while mental effort brings on a slightly larger pupil. But all pupil movements and associated states of arousal are accompanied by a persistent random pupil diameter motion which has previously been attributed to neurological noise, the noise apparently arising in the neurological controller of the pupil reflex control system. Our experiments and signal processing methods show that the amplitude of this pupil noise is an indicator of the sleeping disorder narcolepsy. Narcoleptics are found to have diminished pupil noise amplitudes relative to control subjects. Pupil noise is estimated by statistical procedures which yield unbiased noise measures in the form of six-dimensional Gaussian vectors. Each subject is associated with a Gaussian vector which is optimally projected onto a scalar axis so as to maximize the mean square distance between the narcoleptic and control samples. The Kullback-Leibler discrimination function is estimated and then evaluated for each projection as a means of discriminating narcoleptics from controls. The projected noise measures correctly classify 18 out of 20 subjects. The projected values also form the basis for supporting or rejecting a hypothesis of narcoleptic or control class membership. Parametric and nonparametric hypothesis tests suggest that, with probabilities close to one, narcoleptics and controls are distinguishable classes. To emphasize the importance of pupil noise as a diagnostic tool we present evidence from the neurophysiology literature indicating that Alzheimer's disease and narcolepsy have some of the same brainstem nuclei implicated. Further, Alzheimer patients and narcoleptics share some of the same disturbed sleep patterns.
瞳孔对光反射长期以来一直能够指示精神觉醒状态，范围从困倦到集中的认知努力。这些精神状态通常是根据瞳孔直径或瞳孔面积相对于某个参考值的变化来推断的；例如，困倦的特征是瞳孔小于平均水平，而精神努力则会使瞳孔略大。但是，所有瞳孔运动和相关的觉醒状态都伴随着持续的随机瞳孔直径运动，这种运动以前被归因于神经噪声，这种噪声显然产生于瞳孔反射控制系统的神经控制器中。我们的实验和信号处理方法表明，这种瞳孔噪声的幅度是发作性睡病睡眠障碍的一个指标。发现发作性睡病患者相对于对照组受试者的瞳孔噪声幅度减小。瞳孔噪声是通过统计程序估计的，该程序以六维高斯向量的形式产生无偏噪声测量值。每个受试者都与一个高斯向量相关联，该向量被最佳地投影到一个标量轴上，以便最大化发作性睡病患者和对照样本之间的均方距离。估计库尔贝克-莱布勒判别函数，然后针对每个投影进行评估，作为区分发作性睡病患者和对照组的一种方法。投影噪声测量值正确地将20名受试者中的18名进行了分类。投影值也构成了支持或拒绝发作性睡病患者或对照组成员假设的基础。参数和非参数假设检验表明，发作性睡病患者和对照组在概率接近1时是可区分的类别。为了强调瞳孔噪声作为一种诊断工具的重要性，我们提供了神经生理学文献中的证据，表明阿尔茨海默病和发作性睡病涉及一些相同的脑干核。此外，阿尔茨海默病患者和发作性睡病患者有一些相同的睡眠模式紊乱。
