一种新型结合下肢动觉运动想象和视觉运动想象的脑机接口  

A new BCI by combining kinesthetic motor imagery and visual motor imagery involving lower limbs

作　　者：董煜阳 龚安民 丁鹏 袁密桁[1,2] 王东庆 伏云发 Dong Yuyang;Gong Anmin;Ding Peng;Yuan Miheng;Wang Dongqing;Fu Yunfa(School of Information Engineering and Automation,Kunming University of Science and Technology,Kunming,650500,China;Brain Cognition and Brain Computer Intelligence Integration Group,Kunming University of Science and Technology,Kunming,650500,China;College of Information Engineering,Engineering University of PAP,Xi'an,710086,China)

机构地区：[1]昆明理工大学信息工程与自动化学院,昆明650500 [2]昆明理工大学脑认知与脑机智能融合团队,昆明650500 [3]武警工程大学信息工程学院,西安710086

出　　处：《南京大学学报（自然科学版）》2022年第3期460-468,共9页Journal of Nanjing University（Natural Science）

基　　金：国家自然科学基金(61763022,8217072254,81470084,61463024)。

摘　　要：基于运动想象(Motor Imagery,MI)的脑-机接口(Brain-Computer Interface,BCI)是一类重要的BCI,传统的MI方式是动觉运动想象(Kinesthetic Motor Imagery,KMI),较少采用视觉运动想象(Visual Motor Imagery,VMI).提出一种KMI与VMI混合的BCI并评估其性能.共招募12名被试参加离线与在线实验EEG(Electroencephalogram)数据采集,离线实验先以KMI方式分别进行屈膝和伸膝,然后分别以KMI,VMI和VKMI三种方式行走,由离线分类精度与三种不同方式想象行走的脑激活程度确定在线实验方案.提取EEG幅值包络线特征,并采用朴素贝叶斯分类器、二次线性判别和决策树进行在线分类,验证系统性能.溯源分析表明,使用新的想象方式进行行走想象时,运动皮层的激活时长高于VMI,体感皮层的激活时长高于KMI,混合的想象方式可能更有利于促进这些脑区的可塑性. 12名被试在线测试三分类的平均准确度达到63.29%±0.09%,平均卡帕系数为0.45±0.13.该研究可望为未来研发下肢运动功能障碍康复训练BCI系统提供思路.Brain-computer interface(BCI) based on motor imagery(MI) is an important type of BCI. Kinesthetic motor imagery(KMI) is adopted in the traditional MI,but visual motor imagery(VMI) is rarely used. The current study proposes a hybrid paradigm of KMI and VMI for BCI,and evaluates its performance. A total of 12 subjects are recruited to participate in the offline and online experiment for EEG(Electroencephalogram)/data collection. In the offline experiment,knee flexion and extension are firstly performed with KMI,and then walking is executed with KMI,VMI and VKMI,respectively. The scheme of the online experiment is determined by the offline classification accuracy and the degree of brain activation induced by three different ways of walking imagery. The EEG amplitude envelope characteristics are extracted,and naive Bayes classifier,quadratic discriminant analysis and decision tree are used for online classification in order to test performance of the system.The source analysis shows that the activation duration of the new imagery in the motor cortex is higher than that of the VMI during walking imagery,and the activation duration of the somatosensory cortex is higher than that of the KMI. The hybrid imagery may be more conducive to promoting the plasticity in these brain areas. The average accuracy of three categories in the online test reaches 63.29%±0.71%,and the average Kappa coefficient is 0.45±0.016. This study is expected to provide ideas for the future development of BCI systems for lower limb motor dysfunction rehabilitation training.

关 键 词：视觉运动想象 动觉运动想象 溯源分析 下肢运动意图 在线模型选择 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]