摘要
癫痫疾病发作时,脑电(electroencephalogram,EEG)信号中含有大量的癫痫特征信息,癫痫EEG信号的提取识别和分类研究,对癫痫的预防和治疗具有重大的意义。我们采用经验模态分解(empirical mode decomposition,EMD)算法对发作期、发作间期的EEG进行分解,计算分解后的主要本征模态函数(intrinsic mode function,IMF)分量的波动指数、均值和样本熵值,并组成一组特征向量输入到极限学习机(extreme learning machine,ELM)内进行识别分类。实验结果表明,在需要较少训练样本下,ELM识别分类的准确率达到97%以上。
Epilepsy is a common nervous system disease caused by abnormal discharge of brain neurons.Epilepsy seizure,EEG contains a large number of epileptic features,it has great significance for study on the extraction and recognition of epileptic EEG signals classification for disease prevention.We used empirical mode decomposition(EMD)algorithm of ictal and interictal EEG decomposition,the fluctuation index,mean value and sample entropy of the main IMF components were calculated,the three groups of feature vectors were input into the extreme learning machine(ELM)for identification and classification.The experimental results show that the accuracy of ELM recognition and classification can reach more than 97% with fewer training samples.
作者
宋玉龙
赵冕
郑威
SONG Yulong;ZHAO Mian;ZHENG Wei(College of Electronic Information,Jiangsu University of Science and Technology,Zhenjiang 212003,China)
出处
《生物医学工程研究》
2019年第3期263-268,共6页
Journal Of Biomedical Engineering Research
基金
国家自然科学基金资助项目(61601206)
江苏省自然科学基金资助项目(BK20160565)
江苏省高校自然科学研究资助项目(15KJB310003)
关键词
脑电信号
经验模态分解
本征模态函数分量
样本熵
特征向量
极限学习机
Electroencephalogram signal
Empirical mode decomposition
Intrinsic mode function component
Sample entropy
Feature vector
Extreme learning machine