计算机技术与发展2017，Vol.27Issue(1)：177-181,5.DOI:10.3969/j.issn.1673-629X.2017.01.040
基于脑电信号特征提取的睡眠分期方法研究
Research on Sleep Staging Method Based on Feature Extraction of EEG
摘要
Abstract
Researches on sleep staging is not only the basis of diagnosing sleep related diseases,but also the precondition of sleep quality evaluation,which has vital significance. A new method to extract EEG features is proposed which effectively improves the accuracy of sleep staging. Different from traditional automatic sleep staging method,sleep stage is classified every 30 seconds and time slice for feature extraction is respectively divided into 30 seconds,90 seconds,150 seconds and 210 seconds to study characteristic parameters of difference time slices on the accuracy of sleep stage. Besides,a random forest classifier in Weka tools is adopted to identify sleep state. The result shows that putting wavelet packet coefficients obtained by the 210 s time slice,the permutation entropy from the 30 s time slice and the Petrosian fractal dimension from 90 s time slice as the parameters of the automatic sleep staging,it can get accuracy of 85%,while three kinds of parameters in 30 s time slice above can only reach accuracy of 65%.关键词
睡眠分期/脑电信号/小波包系数/排列熵/Petrosian分形维数Key words
sleep stage/EEG/wavelet packet coefficient/permutation entropy/Petrosian fractal dimension分类
信息技术与安全科学引用本文复制引用
李斐,马千里..基于脑电信号特征提取的睡眠分期方法研究[J].计算机技术与发展,2017,27(1):177-181,5.基金项目
国家自然科学基金资助项目(61201029) （61201029）
