%0 Journal Article
%T 应用深度神经网络对多导睡眠图的睡眠分期研究<br> Application of Deep Neural Network to Study the Sleep Stage Scoring on the Polysomnography
%A 王抒伟
%A 徐富献
%A 钱镶钰
%A 胡桓
%A 何情祖
%A 林海
%A 帅建伟
%J Biophysics
%P 11-25
%@ 2330-1694
%D 2019
%I Hans Publishing
%R 10.12677/BIPHY.2019.72002
%X <div style=\"text-align:justify;\">
	<span style=\"font-size:14px;\">传统上，自动睡眠分期是一项非常具有挑战性且费时费力的任务。大多数现有的自动睡眠分期方法都基于单通道的脑电(electroencephalography, EEG)数据，然而，这些方法忽略了医师从整体上观测多个通道EEG信号进行睡眠阶段评分的过程。为了解决这一问题，我们优化了数据结构，对医师的评分过程进行了详细的学习与建模，提出了一种基于多通道脑电图的自动睡眠评分方法。我们介绍了在原始EEG与EOG样本上使用深度卷积神经网络(convolutional neural network, CNN)进行睡眠阶段评分的监督学习。该网络具有11层，每30 s的睡眠数据作为一个分期，并且不需要任何信号预处理或特征提取。本文使用来自福建省某医院的EEG与EOG及专家评估的多导睡眠图(polysomnography, PSG)数据对系统进行训练和评估。实验结果表明，在自动睡眠分期的研究中不应该忽略EOG数据。我们的系统性能与中级睡眠分期专家的结果相当。<br />
In the field of medical informatics, the automatic sleep staging is a
challenging and time-consuming task, and most existing automatic sleep staging
methods are based on single channel electroencephalography (EEG) data. However,
these methods ignore the physician’s overall observation of multiple channel
EEG and EOG signals for the sleep stage scoring. To resolve this problem, we
propose an automatic sleep scoring method based on multi-channel EEG, including
three-channel EEG and two-channel Electrooculogram (EOG) data. We introduce the
use of a deep convolutional neural network (CNN) on raw EEG samples for
supervised learning of sleep stage prediction, which does not require any
signal preprocessing or feature extraction. We use the EEG and EOG of
polysomnography (PSG) data which have been assessed by medical expert from a
Hospital of Fujian Province to train and evaluate our system. Comparing with
the staging result with single-channel EEG data, we indicate that the EOG data
should not be ignored for a better sleep staging. It shows that the performance
of our system is comparable to that of mid-level experts.</span><span style=\"font-size:14px;\"></span> 
</div>
%K 睡眠分期，多通道，卷积神经网络，脑电图，眼电图<br> Sleep Stage Classification
%K Multichannel
%K Convolutional Neural Network
%K EEG
%K EOG
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=30667