睡眠脑电信号处理和智能判读脑机交互系统研究

EEG signal contains the information related to the functional state of human body, which is an important basis for the analysis of sleep process. However, the traditional sleep interpretation is mainly based on the clinical disease diagnosis and cannot evaluate the sleep problems of sub-healthy people. Especially, the existing signal processing and automatic interpretation methods had limitations for actual sleep data processing. . The significant work in this proposal will focus on the problem of sleep healthy by analyzing and processing the sleep EEG. The key issues are the feature extraction, optimization, and intelligent interpretation. We shall establish the mulit-spatial distribution for different patterns of sleep processes, and provide the evaluation method of sleep phases. Brain-machine interaction control rules are designed for the interface between sleep status (brain) and environment devices (machine) and realize the real-time intelligent interpretation of sleep processes. . The research outcome from this project will be significant for the evaluation of the seriously human sub-health problem. It also extends the application field of brain computer interface to enrich brain science theory.

脑电信号中蕴含了与人体机能状态密切相关的信息，是分析睡眠过程的重要依据。传统睡眠判读以临床疾病诊断为主，缺乏针对亚健康人群睡眠问题的评价方式，而已有睡眠信号处理和判读方法，存在信号特征与判读模式之间相关程度较弱等问题，不能有效用于实际睡眠数据的处理。. 本项目面向亚健康群体的睡眠健康问题，以睡眠脑电信号为研究对象：研究睡眠脑电信号处理方法，重点解决睡眠脑电特征提取和优化问题；研究睡眠过程自动判读方法，建立多类别事件对应的特征参数多维空间分布，给出睡眠过程系统化的评价方式；研究脑机交互控制策略，建立睡眠状态（脑）与环境设备（机）之间的交互接口，构建实时智能化睡眠判读系统。. 项目的实施将为日益严重的人体亚健康问题提供客观评价方式，同时也将拓宽脑机交互智能系统应用领域，丰富脑科学理论方法。

睡眠是重要生理活动，与人身心健康息息相关。针对睡眠脑电信号的特征和睡眠分期的规律性变化，本项目主要研究睡眠脑电的信号处理与特征提取方法、睡眠状态的自动判读方法、以及智能判读交互系统。首先，信号处理和特征提取是重要的前期处理步骤，从数据增强、特征提取、特征优化上对现有方法进行了研究和改进，能够为后续处理提供有用特征样本；其次，在自动判读方法上重点研究了贴近信号特征分布特点并融合人工判读专家模式的机器学习和深度学习模型，同时探索了基于睡眠分期周期性规律的自动修正方法，能够为实际应用提供有效且可行的方案；然后，在智能判读交互系统的研究上，探索功能感官与睡眠状态的关联，进行了智能化睡眠脑机交互系统的实验和研究工作。项目的研究工作将为睡眠医学提供客观的辅助评价方式，同时也能够促进将脑机交互智能化技术拓展到睡眠健康领域的应用中。