基于脑电熵参数的视觉注意力分级研究

发布时间：2019-02-21 20:04

【摘要】： 视觉注意一直是心理学、神经生理学的重要研究领域,近年来随着脑功能探测技术的发展,对于视觉注意力的研究已经从早期的定性描述发展到利用各种神经生理学信号对大脑不同状态下的注意力水平进行定量化分析的阶段。目前,视觉注意力水平的定量化研究尚在选取表征注意力的特征参数与量化的分级方面存在一定的缺陷。寻找新的能客观准确地反映视觉注意力水平的特征参数,建立能稳定、可靠地监测并可动态调控操作者视觉注意力水平的生物反馈系统,将不仅有利于揭示人脑中视觉注意力的心理学、神经生理学方面的作用机制,还将有益于与注意力相关脑神经疾病的诊断和治疗,促进脑-机接口等神经工程新技术及在康复医学中的应用。本文设计了两种用于测试多级视觉注意力水平的脑电反馈实验,通过要求受试者注视屏幕动画并按照提示进行不同强度的想象动作过程并记录多个注意集中程度下的脑电数据。分别采用脑电信号的近似熵、样本熵、多尺度熵、δ-样本熵等基于序列复杂性测度的非线性动力学参数来表征受试者的视觉注意集中程度。经14名受试者实验及其脑电信号熵参数统计分析结果表明,在绝大多数额叶导联(F3、F4、F7、F8、Fz)和部分颞叶导联(T3、T4等)处的多个注意集中程度之间存在显著性差异,并且随受试者注意集中程度的降低,其熵值呈下降趋势;发现样本熵在区分多个注意集中程度时的敏感度最高。为实现多级注意力水平的量化分级,本文优化设计了支持向量机分类器,对注意力水平进行了分级识别,识别率达到85.24%,可以将多种注意力集中程度区分开。最后对受试者注意力集中程度与其想象动作电位特征之间的相关性进行了初步研究。以上基于脑电熵参数的视觉注意力分级研究方法和结果可为脑与认知神经科学基础研究、脑电生物反馈训练以及在线脑-机接口系统设计提供参考依据与技术支持。
[Abstract]:Visual attention has been an important research field in psychology and neurophysiology. In recent years, with the development of brain function detection technology, The study of visual attention has developed from the early qualitative description to the use of various neurophysiological signals for quantitative analysis of attention levels in different states of the brain. At present, the quantitative study of visual attention level still has some defects in the selection of characteristic parameters and quantification classification. To find a new characteristic parameter that can objectively and accurately reflect visual attention level, and to establish a biofeedback system that can monitor and dynamically regulate the visual attention level of the operator, which can be monitored stably, reliably and dynamically. It will not only help to reveal the psychological and neurophysiological mechanisms of visual attention in the human brain, but will also be beneficial to the diagnosis and treatment of attention related neurological diseases. To promote new neural engineering technology such as brain-computer interface and its application in rehabilitation medicine. In this paper, two kinds of EEG feedback experiments are designed to test the multilevel visual attention level. The subjects are asked to look at the screen animation and perform different intensity imaginary movements according to the cues, and record EEG data at multiple attention concentration levels. The nonlinear dynamic parameters, such as approximate entropy, sample entropy, multi-scale entropy and 未 -sample entropy, are used to characterize the visual attention concentration of the subjects. The experimental results of 14 subjects and the statistical analysis of the entropy parameters of EEG showed that in the vast majority of frontal lobe leads (F3F4F4F7F7F8FZ) and part of temporal lobe leads (T3), There was a significant difference in the degree of multiple attention concentration at T4 et al. The entropy showed a decreasing trend with the decrease of the attention concentration of the subjects. It was found that sample entropy had the highest sensitivity in distinguishing multiple attention levels. In order to realize the quantitative classification of multi-level attention level, this paper optimizes the design of support vector machine classifier, and classifies the attention level with a recognition rate of 85.24, which can distinguish the degree of attention concentration. Finally, the correlation between attention concentration and the characteristics of imaginary action potential was studied. The above methods and results can provide reference and technical support for the basic research of brain and cognitive neuroscience, the training of EEG biofeedback and the design of online brain-computer interface system.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2009
【分类号】：R338

【相似文献】