A decision support system for automated identification of sleep stages from single-channel EEG signals
A decision support system for automated identification of sleep stages from single-channel EEG signals
复制标题
DOI：
10.1016/j.knosys.2017.05.005
复制
发表时间：
2017-07-15
影响因子：
8.8
通讯作者：
Subasi, Abdulhamit
中科院分区：
文献类型：
--
作者：
Hassan, Ahnaf Rashik;Subasi, Abdulhamit
一个用于自动检测睡眠阶段的决策支持系统可以减轻医学专业人员手动注释大量数据的负担，加快睡眠障碍的诊断，并有利于研究。此外，实现低功耗和便携的睡眠监测设备需要可靠和成功的睡眠阶段检测方案。本文提出了一种利用单通道脑电信号进行计算机辅助睡眠分期的方法。首先利用可调Q小波变换(Tqwt)将脑电信号段分解为多个子带。然后从所得到的Tqwt子带中提取四个统计矩。该方案利用Bootstrap聚集(袋装)进行分类。使用直观的、统计的和Fisher准则分析来评估特征生成方案的有效性。此外，还利用出袋误差分析对袋装效果进行了评价。阐述了袋装和Tqwt参数的优化选择。提出的自动睡眠评分方法在基准睡眠-EDF数据库和DREAMS主题数据库上进行了测试。我们的方法在Sept-EDF数据库上达到了92.43%、93.69%、9436%、96.55%和99.75%的睡眠阶段2-6状态分类的准确率。实验结果表明，与现有的睡眠分级算法相比，本文提出的自动睡眠评分技术具有更好的算法性能。此外，该方案在AASM和R&K两种睡眠评分标准下的性能相当好，而且该决策支持系统对睡眠状态REM和非REM 1的识别成功率很高。可以预见，由于该方法只使用一路脑电信号，因此该方法将适合于设备实现，消除了医学专业人员对大量记录进行人工注释的负担，加快了睡眠障碍的诊断。(C)2017爱思唯尔B.V.保留所有权利。
A decision support system for automated detection of sleep stages can alleviate the burden of medical professionals of manually annotating a large bulk of data, expedite sleep disorder diagnosis, and benefit research. Moreover, the implementation of a sleep monitoring device that is low-power and portable requires a reliable and successful sleep stage detection scheme. This article presents a methodology for computer-aided scoring of sleep stages using singe-channel EEG signals. EEG signal segments are first decomposed into sub-bands using tunable-Q wavelet transform (TQWT). Four statistical moments are then extracted from the resulting TQWT sub-bands. The proposed scheme exploits bootstrap aggregating (Bagging) for classification. Efficacy of the feature generation scheme is evaluated using intuitive, statistical, and Fisher criteria analyses. Furthermore, the efficacy of Bagging is evaluated using out-of-bag error analysis. Optimal choices of Bagging and TQWT parameters are explicated. The proposed methodology for automated sleep scoring is tested on the benchmark Sleep-EDF database and DREAMS Subjects database. Our methodology achieves 92.43%, 93.69%, 9436%, 96.55%, and 99.75% accuracy for 2-state to 6 state classification of sleep stages on Sleep-EDF database. Experimental results show that the algorithmic performance of the automated sleep scoring technique presented herein achieves better performance as compared to the state-of-the-art sleep staging algorithms. Besides, the proposed scheme performs equally well for two sleep scoring standards, namely- AASM and R&K. Moreover, the proposed decision support system yields high success rate for identifying sleep states REM and non-REM 1. It can be anticipated that owing to its use of only one channel of EEG signal, the proposed method will be suitable for device implementation, eliminate the onus of medical professionals of annotating a large volume of recordings manually, and expedite sleep disorder diagnosis. (C) 2017 Elsevier B.V. All rights reserved.