脑电源模型及其分类算法的研究
|
摘要
从头皮电极记录到的脑电信号,是脑内神经细胞电生理活动在大脑皮层或头皮表面的总体反映,其中包含了大量的生理与疾病信息,是一种无损伤性的成像技术。脑电对神经生理与神经心理的研究意义重大,对人类最终认识自身的神经活动规律也将产生重要影响。
对脑电正、逆问题的研究是脑电问题中一种重要的研究方法,而源模型是求解脑电逆问题的关键因素之一,模型是对脑电活动源的估计,因此,该模型与真实神经元活动源的接近程度决定了脑电源定位的精确程度。
本文主要完成了以下工作:
1.在正问题方面,本文主要讨论并构建了产生脑电头皮电势的各类偶极子源模型——单偶极子源模型、双偶极源模型、三偶极源模型、线性偶极子源模型以及圆盘偶极子源模型。同时,针对各类偶极子模型,根据脑电正演理论,分别产生了不同头皮电势图,并根据头皮电极的位置进行了数据采样。
2.本文分别利用支持向量机算法和BP算法设计分类器,根据脑电正问题产生的仿真数据对脑电源模型进行分类。试验结果表明:虽然BP算法在对大量测试样本进行分类时的错误率略低于支持向量机算法,但随着测试样本数量的减少,分类错误率也在不断升高,并且其训练速度也大大低于支持向量机算法。
3.最后,本文结合脑电正问题相关算法和支持向量机分类器,设计实现了一个脑电源模型及其分类的仿真应用平台。应用该仿真平台可对脑电源模型进行正确且有效的分类,利用分类所获得的脑电源模型,可使得偶极子定位更为准确,更加接近真实脑电活动源。
Electroencephalogram (EEG) is a set of electrical signals recorded from the scalp, it is the reflection of the nerve cells’electric phenomena in the pallium or scalp, it contains a lot of information about physiology and diseases. It is a scatheless imaging technology. EEG has very important research meanings to neurophysiology and neurology, and it will produce an effect on people knowing their own rules about neural movement.
Study about the EEG forward problem and inverse problem is one of the research techniques about the EEG problem. Source model is the key to resolve the inverse problem and is the estimation of the neural current source. Therefore, the model and the real neural source’s degree of similitude decides the degree of accuracy about the location of the EEG source.
The main works of this paper include:
1. About the forward problem, this paper discussed and constructed some dipole source models which induce the potential on the scalp: one dipole source model, two dipole source model, three dipole source model, linear dipole source model and disc dipole source model. And then, we produced different scalp potential graphs of every dipole source models based on the forward problem theory, and then sampled the scalp potential data.
2. In this paper, we designed two classifiers by used the support vector machine and BP algorithm, and we classified the simulated data produced by the forward problem theory. Testing result shows: although BP algorithm’s error rate is lower than support vector machine’s, its training speed is lower than the support vector machine’s, and the error rate is getting higher when the testing data becoming smaller.
3. In the end, this paper designed and carried out a simulating application platform of EEG source model and classification combining the support vector machine algorithm and The EEG forward problem theory. We can get the EEG source model by using this platform, and it will help us locating the dipole more exactly, and more close to the real EEG source.
对脑电正、逆问题的研究是脑电问题中一种重要的研究方法,而源模型是求解脑电逆问题的关键因素之一,模型是对脑电活动源的估计,因此,该模型与真实神经元活动源的接近程度决定了脑电源定位的精确程度。
本文主要完成了以下工作:
1.在正问题方面,本文主要讨论并构建了产生脑电头皮电势的各类偶极子源模型——单偶极子源模型、双偶极源模型、三偶极源模型、线性偶极子源模型以及圆盘偶极子源模型。同时,针对各类偶极子模型,根据脑电正演理论,分别产生了不同头皮电势图,并根据头皮电极的位置进行了数据采样。
2.本文分别利用支持向量机算法和BP算法设计分类器,根据脑电正问题产生的仿真数据对脑电源模型进行分类。试验结果表明:虽然BP算法在对大量测试样本进行分类时的错误率略低于支持向量机算法,但随着测试样本数量的减少,分类错误率也在不断升高,并且其训练速度也大大低于支持向量机算法。
3.最后,本文结合脑电正问题相关算法和支持向量机分类器,设计实现了一个脑电源模型及其分类的仿真应用平台。应用该仿真平台可对脑电源模型进行正确且有效的分类,利用分类所获得的脑电源模型,可使得偶极子定位更为准确,更加接近真实脑电活动源。
Electroencephalogram (EEG) is a set of electrical signals recorded from the scalp, it is the reflection of the nerve cells’electric phenomena in the pallium or scalp, it contains a lot of information about physiology and diseases. It is a scatheless imaging technology. EEG has very important research meanings to neurophysiology and neurology, and it will produce an effect on people knowing their own rules about neural movement.
Study about the EEG forward problem and inverse problem is one of the research techniques about the EEG problem. Source model is the key to resolve the inverse problem and is the estimation of the neural current source. Therefore, the model and the real neural source’s degree of similitude decides the degree of accuracy about the location of the EEG source.
The main works of this paper include:
1. About the forward problem, this paper discussed and constructed some dipole source models which induce the potential on the scalp: one dipole source model, two dipole source model, three dipole source model, linear dipole source model and disc dipole source model. And then, we produced different scalp potential graphs of every dipole source models based on the forward problem theory, and then sampled the scalp potential data.
2. In this paper, we designed two classifiers by used the support vector machine and BP algorithm, and we classified the simulated data produced by the forward problem theory. Testing result shows: although BP algorithm’s error rate is lower than support vector machine’s, its training speed is lower than the support vector machine’s, and the error rate is getting higher when the testing data becoming smaller.
3. In the end, this paper designed and carried out a simulating application platform of EEG source model and classification combining the support vector machine algorithm and The EEG forward problem theory. We can get the EEG source model by using this platform, and it will help us locating the dipole more exactly, and more close to the real EEG source.
引文
[1] 邹凌,朱善安,张迎春.源于脑电偶极子源定位问题.国外医学生物医学工程分册.2003 年第 26 卷第 3 期:103-108.
[2] 刘晓春,彭仕政,谈荣日.脑电磁场及脑内源定位的研究进展.现代生物医学进展.2006,Vol.6, No.6: 9-12.
[3] Gulrajani,M.Ramesh. Bioelectricity and Biamagnetism. John Wiley & Sons, 1998, INC ISBN 0-471-24852-5 QP341.G84.
[4] F.N. Wilson and R.H.Bayley. The electric field of an eccentric dipole in a homogeneous spherical conducting medium. Circulation. 1950,1:84-92
[5] 尧德中,均匀导电球头模型中任意 2n 积子电位的一般性解析解,中国生物医学工程学报,1998,17(2):97-102
[6] 田源.脑波偶极子源定位的研究:[硕士学位论文].广州:第一军医大学,2002.
[7] 张迎春,邹凌.脑电正问题的模型和算法.生物医学工程学杂志.2004,21(2): 337-339.
[8] Baillet and Garnero, 1997 S. Baillet and L. Garnero, A Bayesian approach to introducing anatomo-functional priors in the EEG/MEG inverse problem, IEEE Trans. Biomed. Eng. 1997,44: 374–385.
[9] 詹望,杨福生,一种建立在有限电阻网络模型上的高分辨脑电图的新算法 2 实际头型
[10] Baillet et al., 2001 S. Baillet, J.C. Mosher and R.M. Leahy, Electromagnetic brain mapping, IEEE Signal Process. Mag. 2001, 18: 14–30.
[11] Zhang, 1995 Z. Zhang, A fast method to compute surface potentials generated by dipoles within multilayer anisotropic spheres, Phys. Med. Biol. 1995, 40: 335–349.
[12] 吴清,脑电信号源数值解法的研究,河北工业大学,[博士学位论文].2001
[13] 刘波,脑电场模型研究及实现,国防科学技术大学,[硕士学位论文].2003
[14] Mingui Sun. An Efficient Algorithm for Computing Multishell Spherical Volume Conductor Models in EEG Dipole Source Localization. IEEE Transactions On Biomedical Engineering, 1997,44(12): 1243-1252.
[15] M.R.Schneider.A Multistage Process for Computing Visual bipolar Source of EEG Discharges from Surface Information.IEEE Trans.Biomed.Eng.,1972,19(1)
[16] R.N.Kavanagh,et al.Evaluation of Methods for Three-Dimensional Location Electrical Sources in the Human Brain.IEEE Trans.Biomed.Eng.,1978,25(5)
[17] R.D.Sidman , et al.The Use of Equivalent Source Models in EP Research and Differential Diagnosis.IEEE Frontiers of Engineering in Health Care,1982:64-70
[18] Y.W’atanabe,Y.Sakai.A Graphic Method for Estimating Equivalent Dipole of Localized EEG Discharge.IEEE Trans.Biomed.Eng.,1984,31(5):435-439
[19] D.J. Majer.,R.D.Sidman.,A New Use of SVD in Bioelectric Imaging of the Brain.SVD and Signal Processing II,R.J.Voccaro(Ed.},Elsevier Science Publishers B.V.,1991:497-512
[20] 王玲玲,周纪芗.常用统计方法.上海:华东师范大学出版社,1998:289-297.
[21] 边肇琪,张学工.模式识别.北京:清华大学出版社,2000,第二版:1-10.
[22] 杨光正,吴岷,张晓莉.模式识别.合肥:中国科学技术大学出版社,2001,第一版:1-6.
[23] 孟媛媛,模糊支持向量机的研究与应用.[硕士学位论文].山东师范大学.2006
[24] 于乐源,支持向量机和分类问题的算法研究.[硕士学位论文].大连理工大学,2005,12
[25] R.M.GGulrajani, et al., Moving Dipole Inverse ECG and EEG Solutions. IEEE Trans. Biomed.Eng.,1984,31(12):903 一 910
[26] 杨铁建.基于支持向量机的数据挖掘技术研究:[硕士学位论文].西安:西安电子科技大学,2005.
[27] Jiswei Han, Micheline Kamber 著,范明,孟小峰译.数据挖掘概念与技术[M].北京:机械工业出版社,2001,8:75-116.
[28] 赵晖,荣莉莉,李晓.一种设计层次支持向量机多类分类器的新方法.计算机应用研究,2006 年第6 期:34-37.
[29] Vapnik V.统计学习理论的本质[M].张学工.北京:清华大学出版社,2000:50-65.Vapnik V. The Nature of Statistical Learning Theory [M]. Hang Xuegong .Beijing: Tshinghua University Press,2000.(in Chinese).
[30] 杜晓东,李岐强.支持向量机及其算法研究.信号处理与模式识别.2005 年第 3 期:37-40.
[31] 常军民.基于多特征分类器融合决策的印鉴识别:[硕士学位论文]. 浙江工业大学,2005.
[32] 唐发明,王仲东,陈锦云.支持向量机多分类算法研究.控制与决策.第 20 卷第 7 期,2005 年 7月:746-754.
[33] Kaibo Duan, S Sathiya Keerthi, Aun Neow Poo. An empirical evaluation of simple performancemeasure for tuning SVM hyperparameters. Technical Report CD-O1-11, Control Division, Dept. of Mechanical Engineering, National University of Singapore, 2001.
[34] Kaibo Duan, S Sathiya Keerthi, Aun Neow Poo.An empirical evaluation of simple erformancemeasure for tuning SVM hyperparameters.Technical Report CD-O1-11, Control Division, Dept.of Mechanical Engineering, National University of Singapore, 2001.
[35] Mingui Sun, Ching-Chung Li, Robert J.Sclabassi. A Hierarchiacl Decision Module Based On Multiple Neural Networks, 1997:238-241.
[36] 王树海.BP 神经网络与支持向量机二级识别手写体数字,[硕士学位论文].北京邮电大学,2007
[37] 周志华,曹存根.神经网络及其应用.北京:清华大学出版社,2004:205-250.
[38] 韩力群,人工神经网络理论、设计及应用——人工神经细胞、人工神经网络和人工神经系统.北京:化学工业出版社,2001,9
[39] 张智星.MATLAB 程序设计与应用. 北京: 清华大学出版社, 2002:1-9.
[40] 张航,黄攀.精通 MATLAB6. 北京: 清华大学出版社, 2002:1-6.
[2] 刘晓春,彭仕政,谈荣日.脑电磁场及脑内源定位的研究进展.现代生物医学进展.2006,Vol.6, No.6: 9-12.
[3] Gulrajani,M.Ramesh. Bioelectricity and Biamagnetism. John Wiley & Sons, 1998, INC ISBN 0-471-24852-5 QP341.G84.
[4] F.N. Wilson and R.H.Bayley. The electric field of an eccentric dipole in a homogeneous spherical conducting medium. Circulation. 1950,1:84-92
[5] 尧德中,均匀导电球头模型中任意 2n 积子电位的一般性解析解,中国生物医学工程学报,1998,17(2):97-102
[6] 田源.脑波偶极子源定位的研究:[硕士学位论文].广州:第一军医大学,2002.
[7] 张迎春,邹凌.脑电正问题的模型和算法.生物医学工程学杂志.2004,21(2): 337-339.
[8] Baillet and Garnero, 1997 S. Baillet and L. Garnero, A Bayesian approach to introducing anatomo-functional priors in the EEG/MEG inverse problem, IEEE Trans. Biomed. Eng. 1997,44: 374–385.
[9] 詹望,杨福生,一种建立在有限电阻网络模型上的高分辨脑电图的新算法 2 实际头型
[10] Baillet et al., 2001 S. Baillet, J.C. Mosher and R.M. Leahy, Electromagnetic brain mapping, IEEE Signal Process. Mag. 2001, 18: 14–30.
[11] Zhang, 1995 Z. Zhang, A fast method to compute surface potentials generated by dipoles within multilayer anisotropic spheres, Phys. Med. Biol. 1995, 40: 335–349.
[12] 吴清,脑电信号源数值解法的研究,河北工业大学,[博士学位论文].2001
[13] 刘波,脑电场模型研究及实现,国防科学技术大学,[硕士学位论文].2003
[14] Mingui Sun. An Efficient Algorithm for Computing Multishell Spherical Volume Conductor Models in EEG Dipole Source Localization. IEEE Transactions On Biomedical Engineering, 1997,44(12): 1243-1252.
[15] M.R.Schneider.A Multistage Process for Computing Visual bipolar Source of EEG Discharges from Surface Information.IEEE Trans.Biomed.Eng.,1972,19(1)
[16] R.N.Kavanagh,et al.Evaluation of Methods for Three-Dimensional Location Electrical Sources in the Human Brain.IEEE Trans.Biomed.Eng.,1978,25(5)
[17] R.D.Sidman , et al.The Use of Equivalent Source Models in EP Research and Differential Diagnosis.IEEE Frontiers of Engineering in Health Care,1982:64-70
[18] Y.W’atanabe,Y.Sakai.A Graphic Method for Estimating Equivalent Dipole of Localized EEG Discharge.IEEE Trans.Biomed.Eng.,1984,31(5):435-439
[19] D.J. Majer.,R.D.Sidman.,A New Use of SVD in Bioelectric Imaging of the Brain.SVD and Signal Processing II,R.J.Voccaro(Ed.},Elsevier Science Publishers B.V.,1991:497-512
[20] 王玲玲,周纪芗.常用统计方法.上海:华东师范大学出版社,1998:289-297.
[21] 边肇琪,张学工.模式识别.北京:清华大学出版社,2000,第二版:1-10.
[22] 杨光正,吴岷,张晓莉.模式识别.合肥:中国科学技术大学出版社,2001,第一版:1-6.
[23] 孟媛媛,模糊支持向量机的研究与应用.[硕士学位论文].山东师范大学.2006
[24] 于乐源,支持向量机和分类问题的算法研究.[硕士学位论文].大连理工大学,2005,12
[25] R.M.GGulrajani, et al., Moving Dipole Inverse ECG and EEG Solutions. IEEE Trans. Biomed.Eng.,1984,31(12):903 一 910
[26] 杨铁建.基于支持向量机的数据挖掘技术研究:[硕士学位论文].西安:西安电子科技大学,2005.
[27] Jiswei Han, Micheline Kamber 著,范明,孟小峰译.数据挖掘概念与技术[M].北京:机械工业出版社,2001,8:75-116.
[28] 赵晖,荣莉莉,李晓.一种设计层次支持向量机多类分类器的新方法.计算机应用研究,2006 年第6 期:34-37.
[29] Vapnik V.统计学习理论的本质[M].张学工.北京:清华大学出版社,2000:50-65.Vapnik V. The Nature of Statistical Learning Theory [M]. Hang Xuegong .Beijing: Tshinghua University Press,2000.(in Chinese).
[30] 杜晓东,李岐强.支持向量机及其算法研究.信号处理与模式识别.2005 年第 3 期:37-40.
[31] 常军民.基于多特征分类器融合决策的印鉴识别:[硕士学位论文]. 浙江工业大学,2005.
[32] 唐发明,王仲东,陈锦云.支持向量机多分类算法研究.控制与决策.第 20 卷第 7 期,2005 年 7月:746-754.
[33] Kaibo Duan, S Sathiya Keerthi, Aun Neow Poo. An empirical evaluation of simple performancemeasure for tuning SVM hyperparameters. Technical Report CD-O1-11, Control Division, Dept. of Mechanical Engineering, National University of Singapore, 2001.
[34] Kaibo Duan, S Sathiya Keerthi, Aun Neow Poo.An empirical evaluation of simple erformancemeasure for tuning SVM hyperparameters.Technical Report CD-O1-11, Control Division, Dept.of Mechanical Engineering, National University of Singapore, 2001.
[35] Mingui Sun, Ching-Chung Li, Robert J.Sclabassi. A Hierarchiacl Decision Module Based On Multiple Neural Networks, 1997:238-241.
[36] 王树海.BP 神经网络与支持向量机二级识别手写体数字,[硕士学位论文].北京邮电大学,2007
[37] 周志华,曹存根.神经网络及其应用.北京:清华大学出版社,2004:205-250.
[38] 韩力群,人工神经网络理论、设计及应用——人工神经细胞、人工神经网络和人工神经系统.北京:化学工业出版社,2001,9
[39] 张智星.MATLAB 程序设计与应用. 北京: 清华大学出版社, 2002:1-9.
[40] 张航,黄攀.精通 MATLAB6. 北京: 清华大学出版社, 2002:1-6.