基于小波变换和Teager能量算子的癫痫脑电自动分类  

Automatic Classification of Epileption Activity in EEG Based on Wavelet Analysis and Teager Energy Operator

作　　者：陈金燕[1] 邹俊忠[1] 张见[1] 

机构地区：[1]华东理工大学信息科学与工程学院,上海200237

出　　处：《生物医学工程学进展》2012年第3期147-151,共5页Progress in Biomedical Engineering

摘　　要：提出一种利用小波变换和能量算子对EEG进行预处理提取癫痫特征信号,进行近似熵估计,对脑电信号进行分类的新方法。首先利用小波分析将EEG信号进行4层分解分成多个子频带,对频率接近棘波的第1,2层小波系数计算非线性能量算子,再对能量算子进行近似熵估计,最后用SVM对EEG信号进行分类。结果表明,该方法对癫痫发作期EEG和正常的EEG分类效果比较理想。This paper proposes an epileptic EEG classification mathod that applies the wavelet transformation the nonlinear energy operator （NEO） and the approximate entropy to distinct the differences between health EEG and epileptic EEG with SVM. Firstly, the EEG signals were decomposed up to 4 levels using wavelet analysis. Then, the NEOs of those detailed coefficients with frequencies being closed to spike were calculated, After that, the obtained NEOs were performed approximate entropy computation. At last, EEG signals were classified with SVM. Experimental results showed that the system performed well for classifying beth epilepsy and normal EEG signals.

关 键 词：癫痫波 小波分析 非线性能量算子 近似熵 SVM 

分 类 号：R742.1[医药卫生—神经病学与精神病学]