基于盲反卷积的脑电信号盲分离研究  

Blind Separation of EEG Based on Blind Deconvolution

作　　者：黄璐[1,2] 王宏[3] 

机构地区：[1]东北大学中荷生物医学与信息工程学院,辽宁沈阳110167 [2]大连海洋大学信息工程学院,辽宁大连116023 [3]东北大学机械工程与自动化学院,辽宁沈阳110819

出　　处：《东北大学学报（自然科学版）》2016年第8期1100-1103,共4页Journal of Northeastern University(Natural Science)

基　　金：国家自然科学基金资助项目(61071057)

摘　　要：采用卷积混合模型描述观测脑电信号(electroencephalogram,EEG),提出一种基于盲反卷积的EEG盲分离方法.结合EEG源成分的独立性确定代价函数,并采用共轭梯度法进行迭代寻优.针对EEG仿真实验数据进行方法验证,采用分离信号与源信号之间的相关系数作为验证指标.实验结果表明,本文方法可以较好地实现EEG盲分离,为EEG信号处理和其他生理信号处理分析提供理论和方法借鉴.The convolution mixture model was adopted to describe the as-observed EEG signals,and then a method for the blind separation of EEG based on blind deconvolution was put forward.The cost function was established based on the independence of EEG sources,and iteration was carried out using conjugate gradient method.The verification was implemented with simulation experiment,adopting the correlation coefficients between separated signals and source signals as the verification indexes.Experimental results showthat the method proposed can achieve blind separation of EEG successfully,providing a theoretical and practical reference for the processing of EEG and other physiological signals.

关 键 词：生理信号 脑电信号 信号处理 盲分离 盲反卷积 

分 类 号：R318[医药卫生—生物医学工程]