认知任务驱动的fMRI脑功能网络构建方法  

Cognitive task driven method for fMRI functional brain network construction

作　　者：张笑非 马莲芳[1] 刘若豪 杨阳 郐弘智[4] 钟宁 Zhang Xiaofei;Ma Lianfang;Liu Ruohao;Yang Yang;Kuai Hongzhi;Zhong Ning(Faculty of Information Technology,Beijing University of Technology,Beijing 100124,China;School of Computer Science,Jiangsu University of Science&Technology,Zhenjiang Jiangsu 212003,China;School of Humanities&Social Sciences,Beijing Forest University,Beijing 100083,China;Dept.of Life Science&Informatics,Maebashi Institute of Technology,Maebashi Gunma 371-0816,Japan)

机构地区：[1]北京工业大学信息学部,北京100124 [2]江苏科技大学计算机学院,江苏镇江212003 [3]北京林业大学人文社会科学学院,北京100083 [4]前桥工科大学生命科学与信息工程系,日本群马371-0816

出　　处：《计算机应用研究》2021年第5期1555-1558,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(61420106005);北京市自然科学基金资助项目(4182005);教育部人文社科基金资助项目(14YJC740030)。

摘　　要：针对脑功能网络的构建受到特定大脑图谱对兴趣点描述准确度及覆盖度的限制,提出了基于认知任务信息和神经影像数据的脑功能网络构建方法。首先计算fMRI体素对于认知任务的敏感度,然后在此基础上选取兼顾分布均衡和去中心化的大脑兴趣点,从而构建任务驱动的脑功能网络。实验通过在人脸情绪识别认知任务相关的梭状回构造任务驱动的脑功能网络,其度中心性、聚类系数、全局效率、局部效率这四个复杂网络指标均优于典型大脑图谱梭状回中兴趣点构成的网络。结果表明,计算得到的大脑兴趣点具有更强的整合性,更适合用于表征特定认知任务下的脑功能网络。Since the construction of functional brain network is limited by the accuracy and coverage of the description of regions of interest in a specific brain atlas,this paper proposed a method of functional brain network construction based on cognitive task information and neuroimaging data.The proposed method first calculated the sensitivity of fMRI voxels to cognitive tasks,and then selected the brain regions of interest that balance both distribution and decentralization on this basis,thereby constructing a task-driven functional brain network.The experiment constructed a task-driven functional brain network in the fusiform gyrus related to facial emotion recognition cognitive tasks.Its four complex network indicators of degree centrality,clustering coefficient,global efficiency,and local efficiency are superior to the ones of the network constructed by typical brain atlas.The results show that the brain regions of interest calculated by this method are more integrated and more suitable for characterizing functional brain networks under specific cognitive tasks.

关 键 词：大脑图谱 功能核磁共振成像 脑功能网络 兴趣点选取 

分 类 号：TP389.1[自动化与计算机技术—计算机系统结构]