基于去趋势波动分析和支持向量机的癫痫脑电分类  

Classification of Epileptic EEG Based on Detrended Fluctuation Analysis and Support Vector Machine

作　　者：蔡冬梅[1] 周卫东[1] 李淑芳[1] 王纪文[2] 贾桂娟[2] 刘学伍[2] 

机构地区：[1]山东大学信息科学与工程学院 [2]山东大学齐鲁医院

出　　处：《生物物理学报》2011年第2期175-182,共8页Acta Biophysica Sinica

基　　金：国家自然科学基金项目(30870666);山东省攻关计划项目(2010GSF10243);山东大学自主创新基金资助课题(2009JC004)~~

摘　　要：癫痫脑电的自动检测与分类对于癫痫患者的诊断治疗具有重要意义。本文提出了一种基于去趋势波动分析（detrended fluctuation analysis,DFA）、波动指数和支持向量机（support vectormachine,SVM）的癫痫脑电分类方法。首先采用DFA方法计算脑电的标度指数,然后对脑电进行4-8 Hz、8-13 Hz、13-30 Hz带通滤波并分别计算波动指数,最后使用SVM分类器实现癫痫脑电波的自动分类。对临床脑电信号的实验表明,该方法具有较强的检测性能,分类准确率达到97%。The automatic seizure detection in electroencephalogram（EEG） is significant for diagnosing epilepsy.This paper proposes a new seizure classification approach based on detrended fluctuation analysis（DFA）,fluctuation index and support vector machine（SVM）.Firstly,the scaling exponents were calculated as features for epileptic EEG.EEG was then filtered by 4-8 Hz,8-13 Hz and 13-30 Hz band-pass filters and fluctuation indiced within different frequency bands were calculated.Lastly,a SVM classifier was used to classify seizure automatically.The experimental results show that the proposed classification algorithm is accurate and efficient with a recognition rate of 97%.

关 键 词：脑电图 去趋势波动分析 癫痫检测 支持向量机 波动指数 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]