一种基于EMD的睡眠脑电图梭形波自动识别方法  

A Novel Approach Based on EMD for Automatically Detecting Spindles from Sleep EEGs

作　　者：杨志华[1] 齐东旭[1] 

机构地区：[1]中山大学信息科技学院,广州510275

出　　处：《北方工业大学学报》2005年第1期1-4,共4页Journal of North China University of Technology

基　　金：国家"九七三"重点基础研究发展规划项目 (G19980 3 0 60 7);国家自然科学基金重点资助项目 (60 13 3 0 2 0 );国家自然科学基金资助项目 (0 3 660 8

摘　　要：提出了一种基于EMD分解的睡眠脑电图梭形波自动识别方法.通过对EEG信号作EMD分解,得到具有很好时频特征的Hilbert谱,对Hilbert谱进行分析处理,可以自动识别梭形波出现的准确时间和持续时间.结果显示,识别准确率高达95 .6% .A novel approach for automatically detecting spindles from sleep EEGs (electroencephalo-graph) is presented in this paper. Empirical mode decomposition (EMD) is employed to decompose a sleep EEG, which is usually typically nonlinear and non-stationary data, into a finite number of intrinsic mode functions (IMF). Based on these IMFs, the Hilbert spectrum H(ω, t) of the EEG can be calculated easily, which provides a high resolution time-frequency presentation. Consequently, an algorithm to detect spindles from a sleep EEG accurately is developed. Experiments show encouraging detection results.

关 键 词：EMD分解 Hilbert谱 自动识别 梭形波 

分 类 号：TN318.8[电子电信—物理电子学]