计算机工程与应用 ›› 2017, Vol. 53 ›› Issue (12): 25-30.DOI: 10.3778/j.issn.1002-8331.1703-0114

• 热点与综述 • 上一篇    下一篇

精神分裂症患者工作记忆EEG功能网络属性分析

孙丽婷1,阴桂梅1,3,谭淑平2,赵艳丽2,张进国2,李  东2,李海芳1   

  1. 1.太原理工大学 计算机科学与技术学院 计算机科学系,太原 030600
    2.北京回龙观医院 精神医学研究中心,北京 100096
    3.太原师范学院 计算机系,山西 晋中 030619
  • 出版日期:2017-06-15 发布日期:2017-07-04

Properties analysis of working memory EEG functional network in schizophrenia

SUN Liting1, YIN Guimei1,3, TAN Shuping2, ZHAO Yanli2, ZHANG Jinguo2, LI Dong2, LI Haifang1   

  1. 1.Department of Computer Science, College of Computer Science and Technology, Taiyuan University of Technology, Taiyuan 030600, China
    2.Center for Psychiatric Research, Beijing Huilongguan Hospital, Beijing 100096, China
    3.Department of Computer Science and Technology, Taiyuan Normal University, Jinzhong, Shanxi 030619, China
  • Online:2017-06-15 Published:2017-07-04

摘要: 精神分裂症与一些认知障碍如信息处理、工作记忆等联系紧密,研究工作是记忆任务中精神分裂症患者与正常人的多通道脑电在各个阶段、各个频段存在哪些显著性差异,可为精神分裂症的诊断提供依据。使用相位锁值(Phase Locking Value,PLV)来量化任意两个电极通道之间的相位同步性,构建脑功能网络的关联矩阵,计算不同稀疏度下脑网络的全局属性以及局部属性曲线下面积,在同一阶段、同一频段下对精神分裂症患者和正常人得到的属性值进行非参数检验,找出差异显著的属性及节点,将对应值作为特征训练SVM分类器,进而将精神分裂症患者和正常人分类。属性分析结果表明,工作记忆任务中[θ]和[α]频段发挥主要作用的脑区集中在右侧额叶区和枕叶区,[γ]频段相关的脑区集中在顶叶区;精神分裂症患者额叶右侧区域与枕叶区电极间[θ、][α]波相关性低于正常人,而其顶叶区电极间[γ]波的相关性高于正常人。

关键词: 工作记忆, 精神分裂症, EEG功能网络属性, 特征频段, 特征脑区

Abstract: Schizophrenia and cognitive impairments such as information processing, Working Memory (WM) is closely linked. The aim of this study is to investigate the significant differences in various frequency bands, in each stage of electroencephalographs (EEGs) between schizophrenia patients and normal controls during visual WM task, providing the basis for diagnosis of schizophrenia. This paper quantizes phase synchronism between two pairs of electrodes by phase locking value, and constructs an incidence matrix. The research computes global properties and the area under the curve of local properties in different sparsity of networks and uses non-parametric test to find the significance properties and nodes as features. In addition, this experiment inputs these features into a support vector machine to distinguish patients and controls. The results show that the right frontal and occipital region play a major role in θ and α bands and the γ-band-related brain regions are concentrated in the parietal lobe region during WM task; and in the corresponding region patient’s θ and α wave correlation between electrodes are lower than normal, but γ wave correlation is higher than normal.

Key words: working memory, schizophrenia, EEG functional network properties, feature frequency band, feature brain region