NREM SLEEP STAGING USING WAVCNS INDEX
NREM SLEEP STAGING USING WAVCNS INDEX
复制标题
DOI：
10.1007/s10877-011-9290-4
复制
发表时间：
2011-04-01
影响因子：
2.2
通讯作者：
Bibian, Stephane
中科院分区：
文献类型：
--
作者：
Agrawal, Gracee;Modarres, Mohammad;Bibian, Stephane
Objective.视觉评分的30秒时期的睡眠数据并不总是足以显示足够的细节睡眠的动态结构。它也容易出现相当大的评分者间和评分者内的变异性。此外，它涉及大量的培训和经验，而且非常繁琐、费时、劳动密集和昂贵。因此，需要自动睡眠分期来克服这些限制。由于自然发生的NREM睡眠和麻醉已经被报道具有各种潜在的神经生理学相似性，基于EEG的麻醉深度监测器已经开始渗透到睡眠研究中。本研究调查了WAV（CNS）指数（在NeuroSENSE麻醉深度监测仪中实现）检测NREM睡眠阶段和清醒状态的能力，以获得完整的夜间PSG数据。方法.从24名青少年中获得的完整的夜间PSG睡眠数据由注册PSG技术专家针对不同的睡眠阶段进行评分。使用WAV（CNS）算法对单个额叶通道进行回顾性分析。非参数描述性统计被用来检查WAVCNS指数和睡眠阶段之间的关系。结果WAV（CNS）指数和NREM睡眠阶段之间存在很强的相关性（rho = 0.9458），WAV（CNS）指数值随着睡眠阶段的增加而降低。此外，如通过WAV（CNS）指数分类的不同NREM睡眠阶段之间没有显著重叠，其能够显著区分（P < 0.001）所有清醒和不同NREM阶段对。结论.这项研究表明，自然NREM睡眠深度的变化敏感地反映了WAV（CNS）指数的变化。因此，WAV（中枢神经系统）指数可以作为一个自动的实时指标，自然睡眠的深度与高时间分辨率，并可能是很大的使用自动睡眠分期在常规/术后睡眠图研究。
Objective. Visual scoring of 30-s epochs of sleep data is not always adequate to show the dynamic structure of sleep in sufficient details. It is also prone to considerable inter- and intra-rater variability. Moreover, it involves considerable training and experience, and is very tedious, time-consuming, labor-intensive and costly. Hence, automatic sleep staging is needed to overcome these limitations. Since naturally occurring NREM sleep and anesthesia have been reported to possess various underlying neurophysiological similarities, EEG-based depth-of-anesthesia monitors have started to penetrate into sleep research. This study investigates the ability of WAV(CNS) index (as implemented in NeuroSENSE depth-of-anesthesia monitor) to detect NREM sleep stages and wake state for full overnight PSG data. Methods. Full overnight PSG sleep data, obtained from 24 adolescents, was scored by a registered PSG technologist for different sleep stages. Retrospective analysis was performed on a single frontal channel using the WAV(CNS) algorithm. Non-parametric descriptive statistics were used to examine the relationship between WAVCNS index and sleep stages. Results. A strong correlation (rho = 0.9458) was found between the WAV(CNS) index and NREM sleep stages, with WAV(CNS) index values decreasing with increasing sleep stages. Moreover, there was no significant overlap between different NREM sleep stages as classified by the WAV(CNS) index, which was able to significantly differentiate (P < 0.001) between all pairs of Awake and different NREM stages. Conclusions. This study demonstrates that changes in the depth of natural NREM sleep are reflected sensitively by changes in the WAV(CNS) index. Hence, WAV(CNS) index may serve as an automatic real-time indicator of depth of natural sleep with high temporal resolution, and can possibly be of great use for automated sleep staging in routine/postoperative somnographic studies.