基于经验模态分解的精细复合多尺度排列熵癫痫脑电信号分类方法  

Classification of epileptic EEG signals based on Empirical Mode Decomposition and refined composite multiscale permutation entropy

作　　者：梁袁泽 张学军[1,2] LIANG Yuanze;ZHANG Xuejun(School of Electronic and Optical Engineering,School of Flexible Electronics(Future Technologies),Nanjing University of Posts and Telecommunications,Nanjing 210023,China;National Joint Engineering Laboratory of RF Integration and Microassembly Technology,Nanjing University of Posts and Telecommunications,Nanjing 210023,China)

机构地区：[1]南京邮电大学电子与光学工程学院、柔性电子(未来技术)学院,南京210023 [2]南京邮电大学射频集成与微组装技术国家地方联合工程实验室,南京210023

出　　处：《智能计算机与应用》2024年第5期44-51,共8页Intelligent Computer and Applications

基　　金：国家自然科学基金(61977039);江苏省研究生科研与实践创新计划项目(KYCX21_0712)。

摘　　要：癫痫是一种常见的脑部疾病,通过脑电图能准确地定位人脑中的致痫区域。文章提出一种基于经验模态分解的精细复合多尺度排列熵的癫痫脑电信号自动检测方法,用于解决区分致痫区和非致痫区的癫痫脑电信号难的问题。首先将原信号分割成多个子信号,并对各子信号进行经验模态分解,然后从分解后的不同经验模态函数中提取精细复合多尺度排列熵特征并利用支持向量机进行分类。通过对癫痫脑电的公共数据集测试,实验结果表明准确率、灵敏度和特异度三个性能指标分别达到90.3%,85.0%和96.0%,ROC曲线下面积达0.98。Epilepsy is a common brain disease,which can accurately locate the epileptic regions in the human brain through EEG.In this paper,an automatic detection method of epileptic EEG signals based on Empirical Mode Decomposition and refined composite multiscale permutation entropy is proposed to solve the problem of distinguishing epileptic EEG signals between epileptic and non-epileptic areas.Firstly,the original signal is divided into multiple sub signals,and each sub signal is subject to empirical mode decomposition.Then,refined composite multiscale permutation entropy is extracted from different decomposed empirical mode functions and classified by Support Vector Machine.For the public data set of epileptic EEG,the final experimental results show that the accuracy,sensitivity and specificity of the three performance indicators reach 90.3%,85.0%and 96.0%respectively,and the product under the ROC curve reaches 0.98.

关 键 词：癫痫 经验模态分解 精细复合多尺度排列熵 支持向量机 

分 类 号：TN911.7[电子电信—通信与信息系统]