北京生物医学工程

基于锁相值和图论的脑功能网络特征提取方法

Applying phase locking value and graph theory for feature extraction of brain functional network

作者：李明爱南琳孙炎珺

单位：北京工业大学信息学部(北京 100124)<p>计算智能与智能系统北京市重点实验室(北京 100124)</p>

关键词：运动想象脑电信号；脑功能网络；图论；特征提取；锁相值

分类号：R318.04

出版年·卷·期（页码）：2019·38·1（15-21）

摘要：

目的为了研究大脑运动想象时脑功能网络的状态变化，并区分运动想象任务，本文提出一种基于锁相值和图论的脑功能网络特征提取方法。方法获取Mu节律和Beta节律的运动想象脑电信号(motor imagery electroencephalography, MI-EEG)，计算任意两导相同节律MI-EEG之间的锁相值，分别构建两个节律的脑功能网络，并提取6种全局网络特征参数，对其归一化处理后进行串行融合获得特征向量。最后以支持向量机作为分类器，采用10折交叉验证法，在BCI Competition III Data Sets IIIa数据集上对两种运动想象任务进行分类。结果相比于其他脑网络特征提取方法，本文方法获得了较高的识别率，最高识别率和平均识别率分别达到100.00%和83.33%。结论从脑功能网络的角度，通过构建Mu节律和Beta节律两个运动节律MI-EEG的脑功能网络，提取多个反映大脑网络整体信息的特征，相对于构建单一运动节律MI-EEG的脑功能网络，提取单个网络特征参数，能够有效改善运动想象任务的识别效果，为MI-EEG信号的特征提取方法提供了一种新的思路。

Objective In order to study the state changes of brain functional network caused by motor imagery and distinguish the motor imagery tasks, a feature extraction method of brain functional network is proposed by using phase locking value and graph theory. Methods The motor imagery electroencephalography (MI-EEG) signals corresponding to Mu rhythm and Beta rhythm are obtained, and the phase locking values between any two channels are calculated on the same rhythm MI-EEG. Then, two brain functional networks of Mu rhythm and Beta rhythm are constructed respectively, and six global network parameters are extracted. After normalizing, these parameters are serially fused to construct feature vectors. Finally, support vector machine (SVM) is used as a classifier to verify the effectiveness of this method by using 10-fold cross validation on the public dataset: BCI Competition III Data Sets IIIa. Results Compared with other feature extraction methods of brain network, this proposed method achieves higher recognition rates, and the highest recognition rate and average recognition rate are 100.00% and 83.33%, respectively. Conclusions From the perspective of brain functional network, by constructing the brain functional networks of Mu rhythm and Beta rhythm MI-EEG, multiple network parameters that reflect the whole information of brain network are extracted. Compared with the method which builds one brain functional network using a single rhythm MI-EEG and extracts individual network parameter, this method can effectively improve the recognition effect of motor imagery tasks, and provides a new idea for the feature extraction method of MI-EEG signals.

参考文献：

[1] Milton J, Small SL, Solodkin A. Imaging motor imagery: methodological issues related to expertise[J]. Methods, 2008, 45(4):336-341.

[2] Holmes P, Calmels C. A neuroscientific review of imagery and observation use in sport[J]. Journal of Motor Behavior, 2010, 40(5):433-445.

[3] 刘铁军, 徐鹏, 余茜，等. 运动想象的脑机制及其在运动功能康复中应用的研究进展[J].生物化学与生物物理进展, 2011, 38( 4):299-304.

Liu TJ, Xu P，Yu Q, et al. The research and progress in the mechanism of motor imagery and its application in motor rehabilitation[J]. Progress in Biochemistry and Biophysics, 2011,38(4):299-304.

[4] Cona F, Zavaglia M, Astolfi L, et al. Changes in EEG power spectral density and cortical connectivity in healthy and tetraplegic patients during a motor imagery task[J]. Computational Intelligence & Neuroscience, 2009:279515.

[5] 施锦河, 沈继忠, 王攀. 四类运动想象脑电信号特征提取与分类算法[J]. 浙江大学学报(工学版), 2012, 46(2):338-344.

Shi JH, Shen JZ, Wang P. Feature extraction and classification of four-class motor imagery EEG data[J]. Journal of Zhejiang University (Engineering Science), 2012,46(2):338-344.

[6] Pfurtscheller G. Event-related synchronization (ERS): an electrophysiological correlate of cortical areas at rest[J]. Electroencephalography & Clinical Neurophysiology, 1992, 83(1):62-69.

[7] Gong A, Liu J, Chen S, et al. Time–frequency cross mutual information analysis of the brain functional networks underlying multiclass motor imagery[J]. Journal of Motor Behavior, 2017,50(3):254-267.

[8] Ghosh P, Mazumder A, Bhattacharyya S, et al. Functional connectivity analysis of motor imagery EEG signal for brain-computer interfacing application[C]//2015 7th International IEEE/EMBS Conference on Neural Engineering. Montpellier, France: IEEE Press, 2015:210-213.

[9] Filho CAS, Attux R, Castellano G. Can graph metrics be used for EEG-BCIs based on hand motor imagery?[J]. Biomedical Signal Processing and Control, 2018, 40:359-365.

[10] Gonuguntla V, Wang Y, Veluvolu KC. Event-related functional network identification: application to eeg classification[J]. IEEE Journal of Selected Topics in Signal Processing, 2016, 10(7):1284-1294.

[11] Santamaria L, James C. Use of graph metrics to classify motor imagery based BCI[C]//2016 International Conference for Students on Applied Engineering. IEEE Press, 2017:469-474.

[12] Shrestha B, Vlachos I, Adkinson J A, et al. Distinguishing Motor Imagery from Motor Movement Using Phase Locking Value and Eigenvector Centrality[C]//2016 32nd Southern Biomedical Engineering Conference. Shreveport, LA, USA: IEEE Press, 2016:107-108.

[13] Quiroga RQ, Kraskov A, Kreuz T, et al. Performance of different synchronization measures in real data: a case study on electroencephalographic signals[J]. Physical Review E, 2002, 65(4):041903.

[14] 汪小帆, 李翔, 陈关荣. 网络科学导论[M]. 北京: 高等教育出版社, 2012.

Wang XF, Li X, Chen GR. Network science: an introduction[M]. Beijing: Higher Education Press, 2012.

[15] Tang J, Scellato S, Musolesi M, et al. Small-world behavior in time-varying graphs[J]. Physical Review E: Statistical Nonlinear, and Soft Matter Physics, 2010, 81(5 Pt 2):055101.

[16] Rubinov M, Sporns O. Complex network measures of brain connectivity: uses and interpretations[J]. NeuroImage, 2010, 52(3):1059-1069.

[17] Stam CJ, Reijneveld JC. Graph theoretical analysis of complex networks in the brain[J]. Nonlinear Biomedical Physics, 2007, 1(1):3.

[18] 佘青山, 昌凤玲, 范影乐,等. 基于邻接矩阵分解的脑电特征提取与分类方法[J]. 传感技术学报, 2012, 25(9):1204-1209.

She QS, Chang FL,Fan YL, et al. Feature Extraction and Classification of EEG Based on Adjacent Matrix Decomposition[J]. Chinese Journal of Sensors and Actuator, 2012, 25(9):1204-1209.

服务与反馈：

【文章下载】【加入收藏】

提示：您还未登录，请登录！点此登录