物理学报Issue(5)：050506-1-050506-8,8.DOI:10.7498/aps.63.050506
基于递归量化分析与支持向量机的癫痫脑电自动检测方法
Automatic detection of epileptic EEG based on recurrence quantification analysis and SVM
摘要
Abstract
Automatic detection and classification of epileptic EEG signals have been a significance method for the clinical diagnosis and treatment of epilepsy. The recurrence quantification analysis (RQA) based on the recurrence plot could visualize the recurrence behaviors of dynamical systems from the nonlinear time series and analysis of the recurrence properties. This paper presents a new feature extraction method for epileptic EEG signals based on the recurrence quantification analysis. Experimental results show that the seizure detection directly based on recurrence quantification analysis features has a higher detection performance; especially the classification accuracy based on the deterministic feature can be up to 90.25%. This paper also combines the RQA features with the variation coefficient and fluctuation index, and then puts the feature vectors into a support vector machine (SVM) to automatically detect the epileptic EEG from EEG recordings. Experimental results shows that the proposed methods could achieve a great classification accuracy of 99%.关键词
递归量化分析/递归图/癫痫脑电/支持向量机Key words
recurrence quantification analysis/recurrence plot/epileptic EEG/support vector machine引用本文复制引用
孟庆芳,陈珊珊,陈月辉,冯志全..基于递归量化分析与支持向量机的癫痫脑电自动检测方法[J].物理学报,2014,(5):050506-1-050506-8,8.基金项目
国家自然科学基金(批准号:61201428,61070130,61173079)、山东省自然科学基金(批准号：ZR2010FQ020, ZR2011FZ003)、山东省优秀中青年科学家科研奖励基金(批准号：BS2009SW003)和中国博士后科学基金(批准号:20100470081)资助的课题@@@@Project supported by the National Natural Science Foundation of China (Grant Nos.61201428,61070130,61173079), the Natural Science Foundation of Shandong Province, China (Grant Nos. ZR2010FQ020, ZR2011FZ003), the Shandong Distinguished Middle-aged and Young Scientist Encourage and Reward Foundation, China (Grant No. BS2009SW003), and the China Postdoctoral Science Foundation (Grant No.20100470081) （批准号:61201428,61070130,61173079）
