面向脑电信号的实时脑皮层功能连通性成像技术研究

Traditional studies of neuroimaging have been proposed to reconstruct the activities and study such activity-specific changes in isolation, however, this isolated study is insufficient. Recent studies on bio-electromagnetic problems have shown that a satisfactory understanding of underlying mechanisms requires performing cortical connectivity analysis. Although neuroimaging techniques are constantly being improved and different methods have been comprehensively studied, most studies mainly focus on localization bias or spatial resolution instead of fully investigating unique real-time cortical functional connectivity imaging characteristics. Thus, a complete and rigorous study of the performance of inverse techniques for real-time cortical functional connectivity imaging is placing increasing demands. This project focuses on EEG-based real-time cortical functional connectivity techniques for monitoring and tracing dynamic changes in cortical functional connectivity between different regions of interest on a subject's cortical surface, not on the subject's scalp surface. This project extends the work of current neuroimaging techniques in two important ways. It investigates both a neuroimaging inverse method for cortical functional connectivity, using a distributed source model, and a system that shows the advantage of taking real-time cortical temporal and spectral characteristics into account when doing neuroimaging. The contribution of this project is implementation of a new system for integrating real-time characteristics into the neuroimaging procedure at the cortical source level. The system will be a promising tool for many practical and potential applications.

传统的脑功能成像技术只能用于研究孤立的神经活动信息。近来，关注于大脑不同区域连通性动态机理的研究持续增多，原因在于这类信息有助于更好地揭示脑皮层区域的功能网络，为脑皮层神经活动的理解提供更全面的信息。基于脑电的脑皮层神经成像技术近年来得到了广泛的发展，但目前的技术通常是针对于非实时的离线分析，既没有对脑皮层神经活动进行实时的在线监测，也没有从实时性要求的角度去分析、设计神经连通性成像算法和系统。本项目旨在开发一个实时脑皮层（而不是颅骨之外的头皮层）功能连通性成像系统，此系统能够在神经源层面上实时检测并追踪脑皮层表面不同感兴趣区连通性的时空演变，同时，此系统的实现将在脑机接口、脑电生物反馈、实时诊断脑部疾病等多个领域有着广阔的应用前景。

本项研究开发了基于区域的方法，将生物电磁源逆问题分析技术与连通性分析方法相结合，用于研究脑皮层空间中活动神经源的起因。研究结果证明所提出的方法能够作为一种有效的非损伤性的工具用于生物电域分析和生物医学应用。特别需要指出的是，由于区域是根据生物医学应用中脑部功能分区标准所制定，因此本套方案所得结果更有利于被理解和解释，并且这项研究结果也更有助于不同受试人员间源连通性分析结果的对比。脑电和脑磁图在多种生物医学应用中通常被认作是非常有价值的技术，但是，此类技术所测量的生物电磁数据往往受到噪声的负面影响，导致活动神经源以及活动区域的连通性分析性能的下降。生物电磁数据的噪声抑制是一项极具挑战的任务，因为从数据中去除噪声的同时还要尽可能地保留有用的信息成分。本部分研究，基于稳健主成分分析技术设计了噪声抑制方法，并将其应用于提升生物电磁域源空间信号的评估。本方案优化了测量的生物电磁数据，将其分解为稀疏的噪声部分和低秩的信息部分，结果显示所提方法明显地提高了脑皮层空间源信号评估的质量。本项研究中的多模框架兼顾了皮层脑电所揭示的生物物理空-时-谱特性和脑磁/脑电。在本研究所述框架中有三套方案用于将生物特性整合到脑磁/脑电源成像当中，系数限制得到了较为保守的结果，指数限制得到了较为极化的结果。混合限制方案基于脑电/脑磁数据的特性，动态平衡了系数限制和指数限制方法，保留了两种方法的优点。本部分研究的贡献是提出了一个具备了整合生物特性信息潜能的脑皮层源成像框架。