基于典型相关分析的模拟失重环境下心理生理变化与脑电关联机制的研究

发布时间：2018-01-19 16:18

本文关键词： 失重头低位卧床实验脑电特征典型相关分析双层统计模型　出处：《兰州大学》2014年硕士论文　论文类型：学位论文

【摘要】：航天员在失重飞行期间,大脑对其适应重力变化、调节生理和心理反应,起着十分重要的控制作用。脑功能状态的改变可能引起人体控制调节功能的紊乱,影响航天员的身体健康和航天任务的顺利完成。在航天医学领域,通过脑功能的研究发现,失重或模拟失重会引起脑电的变化,这些变化可能与生理、心理功能的失调有关。但由于传统失重脑研究综合性的欠缺,以及多模态数据异构性对相关分析带来的困难,目前绝大多数失重脑电改变机制的研究都是来自间接性的推断,很少有来源于数据相关分析的直接证据。在此背景下,本文将典型相关分析方法引入到失重条件下的脑电变化与其他生理、心理变化的关系研究中。面对实际应用情况,将典型相关分析与多层统计模型思想结合,提出了基于典型相关分析的双层线性模型。利用典型相关分析对异构数据的适应性,在—6°头低位卧床模拟失重条件下,从心血管功能、情绪状态、认知能力各方面对脑电的影响进行相关分析。同时,针对脑电信号导极的位置分布,将各方面对脑电的影响在20个脑电电极位置上进行具体的定位建模,确定可能发生功能性或结构性变化的具体脑区域。并采用“K最近邻分类器”来验证典型相关分析建模方法的有效性。通过典型相关建模分析发现,模拟失重时脑电、情绪、认知能力以及心血管功能均表现出一定程度的改变,其中心血管功能、认知能力的改变与脑电的变化显著相关。心血管功能主要与两侧颞区的脑电相关,情绪相关的认知能力与脑的顶枕区相关,而注意以及其他基本认知能力的改变主要与额区和中央区相关。分类验证结果表明通过典型相关分析算法融合后的多模态特征具有更高的分类正确率,证实了典型相关分析方法在本文应用中的可靠性和有效性。
[Abstract]:The astronauts in weightlessness during the flight, the gravity to adapt to changes in brain, regulate the physiological and psychological reactions, plays a very important role. The change of human brain function may control regulation function disorder caused by impact, astronaut's health and the smooth completion of the task space. In the field of space medicine, through the study of brain function found that weightlessness or simulated weightlessness will cause changes in EEG, these changes may be related to physiological, psychological function disorder. But because of the lack of traditional weight-loss comprehensive brain research, and the difficulty of multi-modal data heterogeneity of correlation analysis, the vast majority of the EEG change mechanism of weightlessness are from indirect the inferred that there was little direct evidence from the analysis data.
Under this background, the canonical correlation analysis method is applied to EEG changes in conditions of weightlessness and other physiological and psychological changes in the relationship. In the face of the actual application, the canonical correlation analysis and the idea of combining multilevel statistical model, we proposed a two-level linear model based on canonical correlation analysis. By using canonical correlation analysis of heterogeneous the data in weightlessness condition adaptability, 6 degree head down tilt simulated, from the cardiovascular function, emotional status, analyze the effects of the cognitive ability of the EEG. At the same time, the EEG signal guide position distribution, the influence of various aspects of EEG localization specific modeling in 20 EEG electrode position, determine the specific brain regions may have functional or structural changes. And the validity of canonical correlation analysis and modeling method of K nearest neighbor classifier ".
Through the typical modeling analysis, simulated weightlessness EEG, emotion, cognitive ability and cardiovascular function showed a certain degree of change, which was related to cardiovascular function, cognitive changes and EEG changes. Cardiovascular function is mainly related to EEG on both sides of the temporal region, emotional cognitive ability and brain the occipital region, and attention and other basic cognitive ability change mainly with frontal and central areas. The verification results show that the classification through multi-modal feature fusion algorithm of canonical correlation analysis with the correct classification rate is higher, has confirmed the method of canonical correlation analysis in this paper the application of reliability and validity.

【学位授予单位】：兰州大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：R85

【参考文献】