基于子波变换的癫痫脑电信号检测方法的研究
|
摘要
当前癫痫自动检测方法,通常采用希尔伯特黄变换结合脑电信号变换规律进行检测,易受到噪声的干扰,检测结果存在一定的误差。据此,深入研究基于子波变换的癫痫脑电信号检测方法,依据子波变换检测癫痫脑电信号的原理,采用子波变换对含噪的脑电信号进行去噪后,考虑到癫痫患者发病时,脑电信号里异常特征波导致信号波动幅度较大,采用TQWT小波分解并重构脑电信号,提取重构后的脑电信号里有效值与峰峰值指标构成特征分量,根据特征分量设定正常与发病两种样本,通过支持向量机(support vector machine,SVM)分类器对脑电波信号样本分类,实现患者癫痫脑电信号的准确检测。实验结果表明,所提方法可有效检测癫痫脑电信号,检测灵敏度、特异性和准确率均值分别是98.73%、18.84%、98.87%,适用于癫痫脑电信号检测。
The current automatic detection method for Epilepsy usually adopts Hilbert Huang transform combined with EEG signal transformation law, and is susceptible to noise interference, and the result is not accurate. Based on this, the method of detecting Epileptic EEG signals based on wavelet transform was studied, and the principle of detecting Epileptic EEG signals based on wavelet transform was adopted. The wavelet transform was used to denoise the noisy EEG signals, considering the incidence of epilepsy patients. When the abnormal characteristic wave in the EEG signal causing the signal fluctuation amplitude large, the TQWT wavelet was used to decompose and reconstruct the EEG signal, and the eigenvalue and peak-to-peak value of the reconstructed EEG signal were used to form the characteristic component, which was set according to the characteristic component. Both normal and onset samples were used to classify brainwave signal samples by support vector machine(SVM) classifiers to achieve accurate detection of epilepsy EEG signals. The experimental results show that the proposed method can effectively detect epileptic EEG signals. The sensitivity, specificity and accuracy of detection are 98.73%, 18.84%% and 98.87%, respectively. It is suitable for EEG detection of epilepsy.
The current automatic detection method for Epilepsy usually adopts Hilbert Huang transform combined with EEG signal transformation law, and is susceptible to noise interference, and the result is not accurate. Based on this, the method of detecting Epileptic EEG signals based on wavelet transform was studied, and the principle of detecting Epileptic EEG signals based on wavelet transform was adopted. The wavelet transform was used to denoise the noisy EEG signals, considering the incidence of epilepsy patients. When the abnormal characteristic wave in the EEG signal causing the signal fluctuation amplitude large, the TQWT wavelet was used to decompose and reconstruct the EEG signal, and the eigenvalue and peak-to-peak value of the reconstructed EEG signal were used to form the characteristic component, which was set according to the characteristic component. Both normal and onset samples were used to classify brainwave signal samples by support vector machine(SVM) classifiers to achieve accurate detection of epilepsy EEG signals. The experimental results show that the proposed method can effectively detect epileptic EEG signals. The sensitivity, specificity and accuracy of detection are 98.73%, 18.84%% and 98.87%, respectively. It is suitable for EEG detection of epilepsy.
引文
[1]张涛,陈万忠,李明阳.基于AdaBoost算法的癫痫脑电信号识别[J].物理学报,2015,64(12):423-429.
[2]牛宝东,马尽文.基于希尔伯特黄变换的癫痫自动检测[J].信号处理,2016,32(7):764-770.
[3]夏德玲,孟庆芳,牛贺功,等.基于Lempel-Ziv复杂度和经验模态分解的癫痫脑电信号的检测方法[J].计算物理,2015,32(6):709-714.
[4]野梅娜,李艳艳,杨陈军,等.非平衡数据处理方法在癫痫发作检测中的应用[J].西北大学学报(自然科学版),2016,46(6):789-794.
[5]邓赵红,陈俊勇,刘解放,等.面向癫痫脑电图信号识别的径向基最小最大概率分类树[J].电子与信息学报,2016,38(11):2848-2855.
[6]胡世丽,王星光,王观石.子波传过软弱夹层的波形变化规律[J].地球物理学进展,2015,30(4):1896-1902.
[7]Han M,Sun Z.Epileptic EEG signals classification based on wavelet transform and AdaBoost extreme learning machine[J].Anatomischer Anzeiger,2015,165(4):263-8.
[8]李冬梅,张洋,杨日东,等.经验模式分解与代价敏感支持向量机在癫痫脑电信号分类中的应用[J].生物医学工程研究,2017,36(1):33-37.
[9]Das A B,Bhuiyan M I H,Alam S M S.Classification of EEG signals using normal inverse Gaussian parameters in the dual-tree complex wavelet transform domain for seizure detection[J].Signal Image & Video Processing,2016,10(2):259-266.
[10]刘倩倩,戴加飞,李锦,等.基于多变量符号转移熵的癫痫脑电分析[J].数据采集与处理,2016,31(5):983-988.
[11]谢耘,顾建军,普杰信.关于脑电信号准确提取仿真研究[J].计算机仿真,2017,34(1):364-367.
[12]郝崇清,王志宏.基于复杂网络的癫痫脑电分类与分析[J].山东大学学报(工学版),2017,47(3):8-15.
[13]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.
[14]杨昌健,邓赵红,蒋亦樟,等.引入迁移学习的癫痫EEG信号自适应识别[J].计算机工程,2015,41(6):158-164.
[15]姚晓娟,陈旨娟,毓青,等.脑电图-功能磁共振成像技术对局灶性癫痫致痫灶定位的价值[J].中华医学杂志,2015,95(13):987-990.
[2]牛宝东,马尽文.基于希尔伯特黄变换的癫痫自动检测[J].信号处理,2016,32(7):764-770.
[3]夏德玲,孟庆芳,牛贺功,等.基于Lempel-Ziv复杂度和经验模态分解的癫痫脑电信号的检测方法[J].计算物理,2015,32(6):709-714.
[4]野梅娜,李艳艳,杨陈军,等.非平衡数据处理方法在癫痫发作检测中的应用[J].西北大学学报(自然科学版),2016,46(6):789-794.
[5]邓赵红,陈俊勇,刘解放,等.面向癫痫脑电图信号识别的径向基最小最大概率分类树[J].电子与信息学报,2016,38(11):2848-2855.
[6]胡世丽,王星光,王观石.子波传过软弱夹层的波形变化规律[J].地球物理学进展,2015,30(4):1896-1902.
[7]Han M,Sun Z.Epileptic EEG signals classification based on wavelet transform and AdaBoost extreme learning machine[J].Anatomischer Anzeiger,2015,165(4):263-8.
[8]李冬梅,张洋,杨日东,等.经验模式分解与代价敏感支持向量机在癫痫脑电信号分类中的应用[J].生物医学工程研究,2017,36(1):33-37.
[9]Das A B,Bhuiyan M I H,Alam S M S.Classification of EEG signals using normal inverse Gaussian parameters in the dual-tree complex wavelet transform domain for seizure detection[J].Signal Image & Video Processing,2016,10(2):259-266.
[10]刘倩倩,戴加飞,李锦,等.基于多变量符号转移熵的癫痫脑电分析[J].数据采集与处理,2016,31(5):983-988.
[11]谢耘,顾建军,普杰信.关于脑电信号准确提取仿真研究[J].计算机仿真,2017,34(1):364-367.
[12]郝崇清,王志宏.基于复杂网络的癫痫脑电分类与分析[J].山东大学学报(工学版),2017,47(3):8-15.
[13]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.
[14]杨昌健,邓赵红,蒋亦樟,等.引入迁移学习的癫痫EEG信号自适应识别[J].计算机工程,2015,41(6):158-164.
[15]姚晓娟,陈旨娟,毓青,等.脑电图-功能磁共振成像技术对局灶性癫痫致痫灶定位的价值[J].中华医学杂志,2015,95(13):987-990.