北京生物医学工程

基于脑电时-空特征的深度学习失眠障碍检测算法

Deep learning insomnia disorder detection algorithm based on EEG time-space characteristics

作者：范艺晶罗涛李剑峰杨子贤

单位：北京邮电大学（北京 100876） 通信作者：罗涛。E-mail：tluo@bupt.edu.cn

分类号：R318.04

出版年·卷·期（页码）：2022·41·2（161-166）

摘要：

目的现有失眠障碍检测算法一般包括睡眠分期和失眠障碍识别两个阶段，存在差错传播问题，且计算量大。基于此，论文提出一种基于CNN-BiLSTM的深度学习算法，直接检测失眠障碍。方法首先结合睡眠脑电信号时空模式，根据电极分布构造特征矩阵，再通过CNN表达其高级特征。随后馈送至BiLSTM中挖掘睡眠阶段间的时序信息，实现失眠障碍的直接检测。最后按照6:2:2的比例随机设置训练集、验证集、测试集，用准确率作为指标评估算法模型的分类效果。结果在ISRUC-Sleep公开数据集上进行实验，测试集准确率为93.25%，可达到两阶段方法的准确率水平。结论本文设计的CNN-BiLSTM算法模型能够有效检测失眠障碍，将为辅助医生高效地诊断失眠障碍提供可靠技术方法。

Objective The existing insomnia detection algorithms generally include two stages: sleep staging and insomnia recognition, which have the problem of error propagation and mass computing. Based on this, this paper proposes a CNN-BiLSTM algorithm for direct insomnia detection. Methods Based on the spatiotemporal pattern of sleep EEG signal, we first construct the feature matrix according to the electrode distribution and use CNN to extract advanced features. Then the feature vector is input into BiLSTM to extract time sequence information. So as to realize the direct insomnia detection. Finally, we randomly set the training set, validation set and test set according to the ratio of 6:2:2. The accuracy is used to evaluate the effect of the algorithm model. Results Experiments accuracy on ISRUC-Sleep public dataset is 93.25%, which can reach the accuracy level of two-stage method. Conclusions The CNN-BiLSTM algorithm in this paper can effectively detect insomnia, which will provide a reliable technical method for doctors to diagnose insomnia.

参考文献：

[1]? 顾平,何金彩,刘艳骄,等.中国失眠障碍诊断和治疗指南[C]//中国睡眠研究会、黑龙江省中西医结合学会.中国睡眠研究会东北睡眠工作委员会首届学术年会暨黑龙江省中西医结合学会睡眠分会第二届学术年会会议手册.哈尔滨：中国睡眠研究会,2019:77-86.

[2] 张冰涛. 基于生理信息的睡眠障碍识别方法及关键技术研究[D].兰州:兰州大学,2019.

????Zhang BT. Research on sleep disorder recognition methods and key technologies based on physiological information[D]. Lanzhou: Lanzhou University,2019.

[3] Shahin M, Ahmed B, Hamida S, et al. Deep learning and insomnia: assisting clinicians with their diagnosis[J]. IEEE Journal of Biomedical and Health Informatics,2017,21(6):1546-1553.

[4] ?黄小利. 脑电全脑信号及其在睡眠中的应用[D].重庆：西南大学,2019.

Huang XL. EEG global signal and its application in sleep[D].?Chongqing: Southwest University,2019.

[5] 叶仙,胡洁,田畔,等.基于精细复合多尺度熵与支持向量机的睡眠分期[J].上海交通大学学报,2019,53(3):321-326.

Ye X， Hu J, Tian P, et al. Automatic sleep scoring based on refined composite multi-scale entropy and support vector machine[J]. Journal of Shanghai Jiaotong University,2019,53(3):321-326.

[6] 赵文瑞,李陈渝,陈军君,等.失眠障碍与过度觉醒:来自静息态脑电和睡眠脑电的证据[J].中国科学:生命科学,2020,50(3):270-286.

Zhao WR, Li CY, Chen JJ, et al. Insomnia disorder and hyperarousal: evidence from resting-state and sleeping EEG[J]. Scientia Sinica (Vitae), 2020,50(3):270-286.

[7] Hochreiter S, Schmidhuber J. Long short-term memory [J]. Neural Computation, 1997,9(8):1735-1780.

[8] Khalighi S, Sousa T, Santos JM, et al. ISRUC-sleep: a comprehensive public dataset for sleep researchers[J]. Computer Methods and Programs in Biomedicine, 2016, 124:180-192.

[9] Han H, Wang WY, Mao BH. Borderline-SMOTE: a new over-sampling method in imbalanced data sets learning[M]// ICIC 2005: Advances in Intelligent Computing.?Heidelberg: Springer, Berlin, Heidelberg,?2005,3644: 878-887.

[10]Hinton GE, Srivastava N, Krizhevsky A, et al. Improving neural networks by preventing co-adaptation of feature detectors[J]. Computer Science, 2012,3(4): 212-223.

[11] Yang B, Liu H. Automatic identification of insomnia based on single-channel EEG labelled with sleep stage annotations[J]. IEEE Access, 2020,8:104281-104291.

[12]Li F, Yan R, Mahini R, et al. End-to-end sleep staging using convolutional neural network in raw single-channel EEG[J]. Biomedical Signal Processing and Control,2021,63:102203.

[13]Welch P. The use of fast Fourier transform for the estimation of power spectra: a method based on time averaging over short, modified periodograms[J]. IEEE Transactions on Audio and Electroacoustics,1967,15(2):70-73.

[14]王佳威. 脑电波信号的处理方法与应用[D].北京: 北京邮电大学,?2015.

????Wang JW. Preprocessing methods and applications based on EEG[D]. Beijing: Beijing University of Posts and Telecommunications,?2015.?

服务与反馈：

【文章下载】【加入收藏】

提示：您还未登录，请登录！点此登录