文章摘要

文章信息/Info

Title:: Identification of Epileptic EEG Signals based ?on the Tunable Q-factor Wavelet Transform

作者:: 贺王鹏1; 杨琳2Δ; 王芳2; 黄绍平2; 1. 西安电子科技大学空间科学与技术学院，西安710071；2. 西安交通大学第二附属医院，西安710004

Author(s):: HE Wangpeng1; YANG Lin2; WANG Fang2; HUANG Shaoping2; 1.School of Aerospace Science and Technology, Xidian University, Xi’an 710071, China;2. Second Affiliated Hospital of Xi’an Jiaotong University，Xi’an 710004

Keywords:: Epileptic EEG; Tunable Q-factor wavelet transform; Support vector mahine-fator waveler transform; Feature extraction; Classification

摘要:: 针对癫痫脑电（EEG）信号的识别问题，提出了一种基于可调品质因子小波变换（TQWT）的脑电特征提取方法。首先，利用TQWT将EEG信号进行分解，得到各个小波子波带；然后，根据癫痫异常波对应的频率范围，合理的选择小波子波带进行重构，〖JP2〗提取有效值和峰峰值构成特征分量；最后，采用支持向量机进行分类。将所提出方法应用于癫痫脑电信号的识别中，以德国伯恩大学癫痫研究中心采集的典型脑电数据进行验证。实验分析结果表明，所提出的特征提取方法对正常和癫痫发作期EEG信号的分类准确率可达98%。

Abstract:: For the problem of identifying and classifying epileptic EEG signals， we proposed an effective technique based on the tunable Q-factor wavelet transform (TQWT). Firstly, the TQWT was employed to decompose EEG signal into several wavelet subbands. Then, according to the frequency band of epileptic abnormal waves, the EEG signal was reconstructed adaptively via corresponding TQWT wavelet subbands. The root mean square value and peak-to-peak value indicators were calculated as feature vector. Finally, the support vector machine (SVM) was introduced for classification. Moreover, the proposed method was applied to analyze real EEG data collected from the Epilepsy Research Center, University of Bonn, Germany. The results demonstrate that the proposed technique can effectively detect epilepsy disease from EEG signals with good classification accuracy of 98%.