融合同步EEG-MRI信息的脑网络构建及其在P300脑机制中的应用研究

P300的个体差异很大程度上限制了其在诸多领域的应用。虽然脑区间的动态网络层级结构、多模协变量连接模式等对脑信息加工的重要性已被揭示，但传统相关、相干等方法难以有效刻画相应网络的连接模式，也无法深入挖掘P300个体差异的脑机制。为此，本项目将拓展传统P300网络刻画手段，整合高时间分辨率的EEG、功能相关的MRI与大脑结构形态学特征等成像模态-脑状态的时-空信息，发展融合结构-功能-电生理信息的P300信息加工网络构建新方法，探究P300信息传输速度网络、大尺度协变量网络耦合模式，实现脑信息分化与整合网络模式的有效刻画与挖掘；在此基础上，提取两类网络的相关指标，利用监督学习方法构建P300的多变量预测及分类模型，实现个体行为的精准预测和差异化群体的精确识别，促进其在脑-机接口系统性能提升、用户筛选以及临床疾病诊断等领域的应用，并为之提供更全面的理论支撑与技术支持。

The individual variability of the P300 largely restricts its applications in multiple fields. Although the importance of the dynamic network hierarchical structure and the multi-modal covariate connectivity pattern among brain regions has been identified for brain information processing, the traditional correlation and coherence methods are difficult to effectively measure the connectivity patterns of related networks, which results in the difficulty in deeply uncovering the mechanism underlying the individual differences of the P300. To this end, this project will expand traditional P300 network constructions, by integrating the multi-modal and brain state spatiotemporal information of the high temporal-resolution EEG, functional MRI, and morphological features, to develop P300 information processing network construction methods that fuse the structure-function-electrophysiological information. Then, we will explore the information transmission velocity network and large-scale covariate network of the P300 and further uncover the corresponding network architectures that underlie P300 information segregation and integration. Thereafter, related parameters of both two types of P300 networks will be then extracted, and by applying the supervised learning method, we also build the multi-variable prediction and classification model of the P300, to achieve the accurate prediction of individual behavior and precise identification of distinctive groups. This work aims to improve the efficiency of the P300 in promoting the brain-computer interface controlling, efficient user selection, and clinical disease diagnosis, etc, which will further provide a better neural theoretical basis and technical support for its future applications.