基于实时fMRI解码与脑网络建模的听觉信息认知加工机理研究

Exploring the cognitive processing mechanism for auditory information not only benefits the understanding of the brain’s auditory function, but also helps the diagnosis and treatment of auditory disorders induced by the impairment of central nervous system. This project aims to study the cortical representations of auditory information processing from the view of information decoding, and construct a network model of auditory information processing in the human brain. This study intends to employ the real time functional magnetic imaging (real time fMRI) technology to record the dynamic brain activities involved in auditory cognition online, and the multi-variate pattern analysis method to analyze the brain activation pattern and connectivity pattern underlie the auditory object and spatial location processing, as well as the dynamic changes of these patterns, and then construct the network model of temporal-spatial mapping between the brain activities and the auditory stimuli by using the dynamic causal network analysis. This study uses the real time scanning technology, cognitive decoding method and network modeling theory to reveal the neural mechanism underlie auditory information processing, promoting the understanding of auditory information processing theory, so as to apply it to the cause analysis of auditory disorders, and the search of possible approach for recovering the normal auditory function, meanwhile, this study also provides some theoretical reference for the computer simulation of human auditory system.

探索人类听觉信息认知加工机理不仅有利于揭示大脑听觉功能的奥秘,而且有助于中枢系统功能损伤引起的听觉障碍患者的诊断与治疗。本项目将从信息解码的角度研究听觉信息加工过程的皮层表征，构建人脑听觉信息加工的网络模型。研究中拟运用实时功能磁共振成像(real time fMRI)技术在线记录听觉认知脑活动的动态变化，并结合多变量模式分析方法解码听觉内容和方位信息加工的脑区激活模式、脑区连接模式并计算其动态变化，使用动态因果网络分析方法建立脑活动与听觉刺激之间时空映射的网络模型。本项目主要通过实时观测分析技术、认知解码方法和网络建模思想，揭示人脑听觉信息认知加工的神经机理，促进对听觉信息加工理论的理解，以期用于分析听觉障碍的起因及听觉功能恢复的可能途径，同时也为计算机实现人类听觉的模拟提供一定的理论借鉴。

探索人类听觉信息处理的神经机理不仅有利于揭示大脑听觉功能的奥秘,而且有助于中枢系统损伤引起的听觉障碍患者的诊断和干预。本项目利用具有高空间分辨率的功能磁共振成像(fMRI)技术和具有高时间分辨率的脑电（EEG）技术，对人脑感知处理听觉声音、听觉想象和听觉语音刺激数据进行采集；基于脑区激活模式及脑功能连接特征，结合多变量模式分类方法，分别解码听觉信息处理的关键脑区和脑连接，发现颞上回、颞上沟区域激活模式能够有效表征听觉类别信息，并且具有判别力的脑连接模式跨越左右半球，涉及颞上沟、海瑟回及前额叶等高级脑区。利用Granger因果模型和脑网络实时分析，构建了人脑对听觉信息处理的动态网络交互模型,发现人脑对听觉信息处理存在多阶段、动态分布式网络交互特性。总之，本项目主要通过时空观测分析技术、多变量解码方法和网络建模思想，揭示了人脑对不同听觉刺激信息的处理过程，研究成果为探究听觉障碍的起因及听觉功能恢复的可能途径提供了理论基础，同时也为计算机实现对人类听觉的模拟提供一定的理论借鉴。