脑－机接口是指直接把大脑信号转换成控制信号的信息传输通道，它不依赖于外周神经和肌肉等传输通路。脑－机接口是近年来神经工程研究的热点课题，现有的脑－机接口使用的大脑信号多是头皮脑电信号。本论文介绍了一种利用稳态视觉诱发电位脑－机接口。基本原理是利用SSVEP的频率特征来确定使用者的注视方向。不同的闪烁目标采用不同的频率进行标记，通过不同的注视方向来实现对目标的选择。与现有的其它脑－机接口系统相比，系统具有信号记录简单、使用者不需要训练和传输率高等特点。 同时，系统也还存在以下几方面的不足：1)使用者个体差异太大，通用性有待提高，部分受试者不能操作现有系统。2)系统的舒适性有待改进，长时间的视觉闪烁刺激会造成受试者不适，同时降低了系统性能。3)系统还存在假阳性误操作问题。为了解决以上几个问题，从而提高系统的性能，本论文详细研究了电极位置的选择，视觉刺激频率的选择和频率检测算法。采用多导联脑电和独立分量分析结合的方法来讨论信号采集中电极的安放方法。通过对稳态视觉诱发电位子系统的幅度－频率响应分析验证了高频刺激的可行性。经改进后的脑－机接口系统，受试者的平均信息传输率为42 bits/min，与原系统相比有了较大的提高，也远高于文献中报道的5到25 bits/min。在中国康复中心的临床实验中，在线系统并且成功地应用于截瘫病人环境控制器。
Brain-computer interfaces (BCIs) translate brain signals into a control signal without using muscles or peripheral nerves. Brain-computer interface is an attractive topic in neural engineering research. Most present-day BCIs use noninvasive scalp EEGs as inputs.A method of using steady-state visual evoked potentials (SSVEPs) in BCI research is introduced in the dissertation. It harnesses SSVEPs to determine gaze directions. Some frequency-coded virtual buttons flash on the monitor. The user looks at a button and the system determines the frequency of the photic driving response. The button which matches the frequency is the target. It has advantages such as simple experimental approach, little or no training for users, and high information transfer rate. It also has some drawbacks: limited applicability due to individual difference, visual fatigue and comfortlessness evoked by durative stimuli, and false positive caused by alpha rhythm of spontaneous EEG.To improve system performance of SSVEP-based BCI, we study the selection of electrode position, stimuli frequency band and the algorithm of frequency detection. We propose a method of lead selection for the purpose of signal-to-ratio enhancement. Independent Component Analysis (ICA) is employed to decompose EEGs into SSVEP signal and background noise. Optimal lead is selected by comparing signal correlation and noise correlation between different channels. The improved system has reached an average information transfer rate about 42 bits/min for normal subjects. It has also been successfully applied to an environmental controller for the motion-disabled in the clinical experiments in the Rehabilitation Center of China.
