脑-机接口（BCI）可为大脑提供全新对外信息交互方式，已在神经康复等领域展现强大应用潜力。而现有BCI解码算法受高噪声、非平稳、非线性背景脑电（EEG）干扰，很难识别μV以下极微弱感知诱发电位（EP）或思维事件相关电位（ERP）特征，使BCI系统性能受严重损害并制约其应用效果。项目旨在查明BCI过程中背景EEG干扰演进规律，探索识别思维任务所诱发极微弱目标EP或ERP特征的BCI解码方法，突破当前脑-机信息解码瓶颈，显著提高BCI对外操控能力。课题拟先建立背景EEG多参数时-空演变规律模型，进而设计针对性强、抑制效果好的EEG噪声时-空联合滤波方法，发展能有效识别极微弱EP或ERP特征提取与分类算法，由此搭建脑-机高匹配BCI系统并在临床神经康复应用中初步验证效果。项目研究成果将创新发展μV以下极微弱思维脑电特征解码方法，全面优化BCI整体性能、提升其实用水平，具有重要科学意义与应用价值。

Brain-computer interfaces (BCIs) provide a completely new pathway for the brain to have a direct information exchange with the outside environment. BCIs have shown great potentials in the neural rehabilitation field. However, due to high-noise, non-stationary and non-linear background electroencephalography (EEG), traditional BCI decoding techniques could hardly recognize the miniature evoked potentials (EPs) and event-related potentials (ERPs) smaller than 1 microvolt in amplitude, which severely restrict the BCI’s performance and its capacity for neural rehabilitation. This project aims to investigate the law of evolution process of background EEG during the BCI controlling, develop an advanced BCI decoding method to recognize the miniature EPs and ERPs relevant to the brain intention, and thereby greatly enhance the performance of BCI systems. Firstly, this project will construct a multivariate model to accurately describe the evolution process of background EEG in the time-space domain. On that basis, an elaborate temporal and spatial joint filtering method will be designed to suppress the background EEG noise. Then, a powerful algorithm will be developed for the extraction and recognition of miniature EPs and ERPs. Last, this project will apply the proposed EEG decoding method to a clinic BCI system for neural rehabilitation to examine its effectiven ess. The outcome of this project will advance the decoding technique of miniature EEG features below 1 microvolt in amplitude and promote the performance of BCI system in practical application, which is significant for both scientific and clinic researches.