基于复杂网络的癫痫脑电分类与分析

发布时间：2018-06-09 22:37

本文选题：复杂网络 + 癫痫脑电　；参考：《山东大学学报(工学版)》2017年03期

【摘要】：为提取癫痫发作与间歇期脑电信号的特征,提出利用构建癫痫EEG(electroencephalogram)网络的方法来刻画脑电信号。研究各变量均可测情况下的Lorenz和R9ssler混沌系统,利用其各变量的输出混沌时间序列构建复杂网络,发现构建的复杂网络拓扑图与其混沌吸引子存在形态相似性,说明由时间序列构建的复杂网络能刻画其原信号特征。对于多维系统中仅有一维可测时,多维时间序列由相空间重构得到。利用相空间重构方法对癫痫发作和间歇期脑电信号构建复杂网络进行分析。研究结果表明,癫痫发作时其网络拓扑较间歇期存在明显不同,且其平均路径长度显著增加,而递归率及其波动范围都显著降低,这些网络特性可以用来刻画脑电信号的特征,从而为癫痫疾病的自动辨识与预测提供基础。
[Abstract]:In order to extract the characteristics of epileptic seizures and EEG signals, a method of constructing epileptic EEG (electroencephalogram) network is proposed to characterize EEG signals. The chaotic systems of Lorenz and R9ssler are studied under the conditions of all variables, and complex networks are constructed by using the output chaotic time series of its variables to find the complex network constructed. The topological graph is similar to the chaotic attractor. It shows that the complex network constructed by the time series can depict its original signal characteristics. For only one dimension in the multidimensional system, the multidimensional time series is reconstructed by phase space. The complex network of epileptic seizures and the intermittent period of EEG is constructed by phase space reconstruction. The results show that the network topology of epileptic seizures is significantly different from that in the intermittent period, and the average path length of the epileptic seizures is significantly increased, and the recurrence rate and its fluctuation range are significantly reduced. These network characteristics can be used to characterize the EEG signal and provide the basis for the automatic identification and prediction of epilepsy.
【作者单位】：河北科技大学电气工程学院;天津大学电气与自动化工程学院;
【基金】：河北省自然科学基金资助项目(F2014208013)
【分类号】：O157.5;R742.1

【相似文献】