项目名称: 大脑运动想象系统信息表征提取算法与模式识别研究
项目编号: No.61273361
项目类型: 面上项目
立项/批准年度: 2013
项目学科: 自动化技术、计算机技术
项目作者: 张江
作者单位: 电子科技大学
项目金额: 82万元
中文摘要: 运动想象是脑机接口的主要研究范式,常被用于研究运动执行内在潜意识加工过程。运动想象任务引起脑功能活动产生的是微弱信号,其脑网络信息整合的神经机制仍然不清楚。本项目拟针对运动想象生物反馈对功能网络调控的认知问题,解决利用功能磁共振成像(fMRI)时运动想象脑功能活动信息表征提取算法和模式识别的关键技术问题,主要包括:发展检测fMRI脑功能活动信息的仿射聚类算法和多体素贝叶斯支持向量机分类器模式识别方法,实现运动想象脑网络特征信息有效提取;发展基于图论和小世界网络指标(聚类系数、最短路径和节点度分布)的脑网络分析方法,探测运动想象脑网络信息特征;最后针对运动想象脑网络环路,利用动态脑网络分析方法探测运动想象反馈信息对脑功能网络的调控行为。本项目关于运动想象脑网络的研究成果将为脑机接口提供部分基础理论,同时所发展的脑信息表征提取算法和模式识别方法将为脑功能信息检测提供新的手段。
中文关键词: 功能磁共振成像;运动想象;多体素模式分析;聚类分析;小世界网络
英文摘要: Motor imagery as a chief paradigm for the study of brain-computer interface, is often used to investigate the inner subconscious processing in a motion being implemented. The brain, stimulated by a series of imagery task, produces signals that rather weak, and the neural mechanism governing the integration of brain network information still remains unclear. The present study is thus directed towards the cognitive issues arising from the adjustment of the brain function network as revealed in a motor imagery feedback. We attempt to address the critical technical problems involved in the fMRI-based extraction of information representation from the brain motor imagery system and the related pattern recognition. Our approaches mainly include: developing the affinity propagation clustering method, and the pattern recognition of multi-voxel Bayes-SVMs, both intended for examining fMRI brain functions, in a bid to achieve an effective extraction of of motor imagery features within the cerebral network.; developing the brain network- analyzing method based on graph theory and small world network indexes (clustering coefficient, shortest path, node distribution and degree distribution), exploring the feature information in the brain network in a motor imagery task; and finally, targeting the loop circuit of brain network
英文关键词: Functional magnetic resonance imaging (fMRI);Motor imagery;Multi-voxel pattern analysis (MVPA);Clustering analysis;small-world network