一种通过脑电信号实时检测双眼竞争的方法  

A Method for Real-time Detection of Binocular Rivalry through Electroencephalogram

作　　者：林中林[1] 张雅静[2] 高小榕[2] 张长水[1] 

机构地区：[1]清华大学自动化系,北京100084 [2]清华大学生物医学工程系,北京100084

出　　处：《航天医学与医学工程》2007年第5期381-384,共4页Space Medicine & Medical Engineering

基　　金：国家自然科学基金资助项目(60401008);北京市自然科学基金重点项目(3051001)

摘　　要：目的为实时准确地通过脑电信号分析双眼竞争中的意识交替感知,提供新的客观检测途径。方法实验中,被试者的双眼分别给予闪烁频率略有不同的两幅图片刺激,产生双眼竞争现象,即被试者意识到的是两幅图片的交替转换;同时,被试者的脑电中也会产生和闪烁频率相同的稳态视觉诱发电位成份,该电位随感知交替变化。利用记录的多通道脑电信号,首先用一小段已知转换点的信号,训练两个能够捕捉切换点附近信号特性的检测模板,再用模板在整段信号上滑动,进行典型相关分析,给出其它未知的切换点,由此达到客观检测双眼竞争中主观意识的目的。结果在仿真数据和真实实验数据的分析中,用本文提出方法得到的客观检测结果,和主观报告的匹配程度都好于用已有的谱分析方法。结论这种基于模板检测方法,相比已有的谱分析方法具有更好的抗噪性能,并能够充分利用谐波的信息。在客观检测双眼竞争中具有实用价值。Objective To propose a new approach for real-time detection of binocular rivalry through electroencephalogram （FFG）. Methods The subjects were required to gaze at two seperate images, flickering at slightly different frequencies, with one of his/her eyes, on one image and the other eye on the other image. The perception will alternate between the two images voluntarily （ binocular rivalry）. At the same time, steady-state visual evoked potentials with the same frequency of the flickering images will emerge. Using a segment of multichannel FFG signals, two template models were trained to capture the FFG characteristic at the time of perception alternation. Then, the templates were moved on the whole length of the FFG signal to find out other time points of perception alternation. Results The objectively detected result by our proposed method can match the subjective reports better than currently used method based on power spectral analysis. Conclusion This model based detection method using multichannel signals and harmonic information is more robust than the currently used method based on power spectral analysis.

关 键 词：双眼竞争 实时检测 稳态视觉诱发电位 典型相关分析 

分 类 号：R319[医药卫生—基础医学]