用于轻度认知障碍患者空间认知训练的脑机接口与虚拟现实融合关键技术研究

Cognitive rehabilitation in patients with mild cognitive impairment is a challenging problem in the field of neurological rehabilitation. If not timely intervention and rehabilitation, it is possible to develop into severe cognitive impairment. Among them, the key challenge to be solved is how to improve the effect of the patients' personalized self-adaptive rehabilitation and effectively evaluate it in real time. This challenge transcends the traditional approach. Considering space cognitive defects are the main manifestations of the patients, the project intends to focus on study a new method of spatial cognition training: the brain-machine interface and virtual reality are integrate, and its key technologies are used to personalized and adaptive training of spatial cognitive abilities of patients, and the real-time and effective evaluation of training effect. First of all, it is proposed to design effective EEG preprocessing, feature extraction and classification method, in order to establish a regression prediction model of EEG characteristic value and spatial cognition ability of the patients. Secondly, the effective closed-loop neural feedback training model is constructed, which integrates the BCI and the virtual reality, and personalized adaptive games and training programs were designed. The project is expected to provide intergrating key new technology for the space cognitive training of patients with mild cognitive impairment , also helps to understand the technology playing a role of brain mechanisms, supply this kind of cognitive rehabilitation in patients with effective new method and on the basis of brain science.

轻度认知障碍患者的认知康复是神经康复领域的一个挑战性难题，如不及时干预和康复，有可能发展成为重度认知障碍。其中，急需解决的关键挑战是康复过程中如何提高患者的个性化自适应康复效果并对其进行实时有效的评价。而这种挑战超越了传统方法的解决范围。鉴于空间认知缺陷是该类患者的主要表现，项目拟聚焦研究空间认知训练的新方法：将脑机接口与虚拟现实相融合，并将其关键技术用于对患者空间认知能力的个性化、自适应训练以及对训练效果的实时有效评价。首先，拟设计有效的脑电信号预处理、特征提取与分类方法，建立该类患者脑电信号特征值与空间认知能力的回归预测模型。其次，拟构建有效的闭环神经反馈训练模型，通过该模型融合脑机接口与虚拟现实，设计个性化自适应游戏以及训练方案。项目的开展有望为轻度认知障碍患者空间认知训练提供融合的关键新技术，也有助于理解该技术发挥作用的脑机制，为该类患者的认知康复提供有效的新方法和脑科学依据。

考虑到当前国内外轻度认知障碍疾病无有效康复训练方法的实际背景，项目组成员在国家自然科学基金的资助下，深入开展了脑电信号预处理方法、脑电信号特征提取方法、脑电信号分类方法、脑电信号特征强度和空间认知测评等量表得分的回归预测模型、个性化自适应闭环神经反馈模型与BCI-VR训练系统、空间认知游戏设计与训练基本方案等方面的研究。本项目取得了如下重要结果：提出了高精度脑电信号预处理方法、脑电信号全自动化预处理方法等新型智能预处理方法；研究出了多维排序条件互信息多光谱图像、多维条件互信息共空间模式等空间认知脑电信号特征提取方法；探索出了加权排序条件互信息、特征融合多光谱图像法等轻度认知障碍脑电信号特征提取方法；研究出了多尺度高密集卷积神经网络等空间认知脑电信号特征分类方法；探索出了多视图卷积神经网络等轻度认知障碍脑电信号分类方法；研究出了基于改进的图卷积神经网络的高性能脑电信号分类方法；建立了一个针对脑电信号特征强度与空间认知测评等量表得分的有效多元线性回归模型；设计出了个性化自适应闭环神经反馈模型、BCI-VR闭环神经反馈训练系统；搭建出了6个空间认知训练系统，设计出了水迷宫训练游戏与城市漫游测试游戏、虚拟社区寻址训练游戏、城市寻路训练游戏、虚拟无人机空间认知训练游戏、虚拟空间记忆图案搭建测试游戏、城市驾驶比赛训练游戏以及对应的训练基本方案；设计出了VR游戏玩家脑电信号特征库、MCI认知数字诊疗系统，实现了空间认知脑电信号的高效加密。本项目的研究为空间认知有效训练与评估领域开拓了新视野，本项目的研究成果有望能向其他认知域的训练与评估方面进行应用与拓展，有助于整体提升我国轻度认知障碍的数字诊疗研究水平。

内容获取失败，请点击重试