EEG-NIRS-fMRI融合研究运动及想象对情绪的调节机制与脑机交互应用

It is very important for the people with high anxiety to have emotional health. Aerobic exercise (hereinafter referred to as exercise) can improve negative emotion. However, the brain mechanism of emotion regulation is still not clear. In addition, the traditional methods rarely use imagery strategies to promote the adjustment of emotion, whose brain mechanism of this promoting effect is not clear. Therefore, this project proposes a new paradigm of the effect of exercise and imagery on the adjustment of mood in order to explore the mechanism. The cortical and subcortical structures involved in emotion regulation are more; the traditional methods mainly use single mode brain imaging method to tracking research, however, the single mode method has the limitation of its own spatial resolution and temporal resolution, which provides single brain activity information, moreover, brain signals monitored during exercise and imagery also require unique processing methods. Therefore, this project intends to fuse the EEG-NIRS-fMRI multi -mode brain imaging information, combining the traditional tracking research with short-term dynamic monitoring research, further researching the regulation mechanism of the effect of exercise and imagery on the adjustment of mood based on the new paradigm. Finally, based on the distribution of brain activation areas and brain function network, neural information characteristics of fMRI, NIRS, and EEG related to emotion regulation by exercise and imagery are extracted, brain computer interaction prototype system based on EEG-NIRS-fMRI fusion are established, which are expected to be used by people with high anxiety in order to monitor and evaluate emotion regulation by exercise and imagery strategies.

高焦虑特质人群的情绪健康很重要，有氧运动（以下简称运动）具有改善负性情绪的作用，然而其调节情绪的脑机制依然不清晰。此外，传统方法少有利用想象策略促进情绪的调节，其促进作用的脑机制亦不清楚。为此，本课题提出一种新的运动及想象对情绪调节的实验范式以探索该机制。情绪调节涉及的皮层和皮层下结构较多，传统方法主要采用单模式脑成像追踪研究，然而单模式方法受自身空间或时间分辨率的局限，提供的脑活动信息也单一。因此，本课题拟融合EEG-NIRS-fMRI多模式脑成像信息，结合传统的追踪研究与短时的动态监测研究，基于提出的新实验范式深入研究运动及想象对情绪的调节机制。最后，基于上述揭示的脑区激活分布以及脑功能网络回路，提取运动及想象调节情绪相关的fMRI、NIRS、EEG神经信息特征，构建EEG-NIRS-fMRI融合的脑机交互原型系统，可望用于高焦虑特质人群采用有氧运动及想象策略调节情绪的监测和评估。

个体或群体的情绪健康很重要，特别是情绪处于亚健康的人。有氧运动改善负性情绪的作用机制依然不清晰，并且如何对情绪进行监测、评估和分类。针对以上问题，1)设计新的运动及想象对情绪调节的实验研究范式; 2）基于功能近红外光谱（fNIRS）研究了有氧运动和运动想象调节情绪的作用机制，并进行监测与评价；3）基于脑电（EEG）研究了情绪特征的提取和识别方法；4）构建了基于fNIRS/EEG脑机接口（BCI）的在线情绪监测、评估和调节原型系统；为使BCI系统实用化，研究了BCI人因工程。与运动想象相比，有氧运动20min后，状态-特质焦虑量表（STAI）和心境状态量表（POMS）评分下降，前额叶皮层具有不对称性和偏侧性（左侧在情绪调节中占主导地位）。fNIRS记录的血红蛋白浓度增加与STAI和POMS评分呈负相关，有氧运动对情绪调节有很好的效果，研究表明便携式fNIRS可有效用于监测和评估有氧运动的情绪调节。对于情绪EEG识别，基于可调Q因子小波变换和二元灰狼优化识别了生理信号数据集情感分析数据库中32名被试的五类情感信号样本，经过6倍交叉验证，最大识别准确率为90.48%，灵敏度为70.25%，特异性为82.01%，Kappa系数为0.603。实验结果表明，该方法在识别多种类型的脑电情感信号时具有良好的性能指标，与传统方法相比，具有更好的性能改进。对于情绪EEG特征提取，采用希尔伯特⁃黄变换(HHT)提取情绪EEG特征并利用高斯径向基函数核的SVM取得的平均、最高和最低分类精度分别为90.57%±4.13%，96%和88%，这表明HHT能有效地提取情绪相关EEG特征，采用高斯径向基函数(RBF)核的SVM可以获得较好的识别结果。基于深度信念网络（DBN）和随机森林（RF）探索了跨被试情绪识别模型，DBN和 RF的平均分类精度分别为83.49%±2.6%和81.73%±2.22%，可望为跨被试的情绪识别模型提供参考。考虑到与情绪BCI以及其他BCI的实用化问题，BCI人因工程要求以BCI用户为中心设计和评价BCI系统。课题所得结果有望为进一步研发基于EEG/fNIRS BCI的在线实时情绪监测、评估和调节系统提供思路，可望用于个体或群体的情绪健康。