基于DD-DWT和Log-Logistic参数回归的癫痫脑电自动识别方法
DOI:
作者:
作者单位:

吉林大学通信工程学院长春130012

作者简介:

通讯作者:

中图分类号:

TH79

基金项目:

吉林省科技发展计划自然基金(20150101191JC)、吉林大学研究生创新项目(2016092)、中央高校基本科研业务费专项资金(451170301193)资助


Automatic epilepsy EEG recognition method based on DD-DWT and Log-Logistic parameter regression
Author:
Affiliation:

Department of Communication Engineering, Jilin University, Changchun 130012, China

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    针对现有癫痫脑电(EEG)识别算法分类模式单一、普适性不强的问题,提出了一种新的基于双密度离散小波变换(DDDWT)和LogLogistic参数回归(LLPR)的脑电信号自动识别方法。不仅利用了DDDWT算法的分解特性,还建立了脑电信号的LLPR模型,并将二者有机的结合,从而更好的发挥算法的优势。滤波后脑电信号由DDDWT进行6层分解,提取各子频带系数的小波域能量波形,并结合LLPR模型计算尺度参数α和形状参数β以表征信号,将构成的特征向量送入遗传算法(GA)优化的支持向量机(SVM)得出识别结果,从而实现脑电信号的自动识别。所提方法在处理A\D\E与AB\CD\E两种多模式脑电分类问题时,识别率分别为98.90%和97.75%。实验结果表明,所提算法更符合实际应用需求,可以较好地解决多类脑电信号识别问题,具有良好的普适能力和分类性能。

    Abstract:

    Aiming at the problems of single classification mode and poor universality of existing epilepsy EEG recognition algorithms, a novel EEG signal automatic recognition method is proposed based on DoubleDensity Discrete Wavelet Transform (DDDWT) and LogLogistics parameter regression (LLPR). This method not only utilizes the decomposition capacity of DDDWT algorithm, but also constructs the LLPR model for EEG signal, integrates the two algorithms organically, and fully exploits the advantages of the two algorithms. In this study, the filtered EEG signals are decomposed into six levels with DDDWT, and the wavelet coefficients of various subbands are transformed to the energy waveforms in wavelet domain to acquire the feature parameters using the LLPR model. The scale parameter α and shape parameter β are calculated to characterize the EEG signal. The feature parameters extracted from all the subbands are composed as the eigenvalues, which are fed to support vector machine (SVM) optimized with genetic algorithm (GA) to obtain the final classification results, thus the EEG signal automatic recognition is achieved. When the proposed method was used to deal with two multimode EEG classification problems of A\D\E and AB\CD\E, the satisfied accuracies of 98.9% and 97.75% were obtained respectively. Experiment results indicate that the proposed method can meet the actual application requirement, is more appropriate for solving the recognition problems of multiclass EEG signals, has good universality and classification performance, and has great value in practical applications dealing with epileptics.

    参考文献
    相似文献
    引证文献
引用本文

李明阳,陈万忠,张涛.基于DD-DWT和Log-Logistic参数回归的癫痫脑电自动识别方法[J].仪器仪表学报,2017,38(6):1368-1377

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:
  • 最后修改日期:
  • 录用日期:
  • 在线发布日期: 2017-07-21
  • 出版日期: