基于Chirplet变换的变频视觉诱发电位脑-机接口研究

Chirp Stimuli Visual Evoked Potential Based Brain-Computer Interface by Chirplet Transform Algorithm

脑-机接口(brain-computer interface,BCI)是在大脑与外部设备间建立一个直接的信息交流通路,它无须依赖外周神经肌肉系统而仅通过脑电信号特征提取与模式识别来实现思维表达或指令操作.变频视觉诱发电位(chirp stimuli visual evoked potential,Chirp-VEP)是最近提出的一种脑电诱发新模式,可作为BCI控制信号,极富应用潜力.然而Chirp-VEP的诱发条件、信号处理、特征提取方法等都缺乏充分研究.本文采用不同起始频率和chirp调频率进行了Chirp-VEP诱发实验,利用Chirplet变换(chirplet transform,CT)等4种时频分析方法提取了ChirpVEP信号特征.研究结果表明,相较于其他时频分析方法,CT可获得更高的VEP信噪比与正确识别率.在8名受试者参加的在线BCI测试中,Chirp-VEP的总平均正确识别率高达97.8%,进一步验证了Chirp-VEP应用于BCI控制的潜力.

A brain-computer interface （BCI） is to set up a direct communication pathway between the brain and an external device. It realizes the human thinking expression or command operation through feature extraction and pattern recognition of electroeneephalography（EEG） rather than relying on the per- pheral nerve muscle system. The newly proposed chirp stimuli visual evoked potential（Chirp-VEP） can be used as a most promising BCI control signal. However, the inducing conditions, signal processing, feature extraction methods of Chirp-VEP need comprehensive study. Different initial frequencies and chirp rates were adopted to evoke the Chirp-VEP in this paper. Four kinds of time-frequency analysis methods, such as chirplet transform（CT） and so on were used to extract the time-frequency characteris- tics of Chirp-VEP signals. The results show that compared with the other three methods, CT can obtain higher VEP signal-to-noise ratio and accuracy recognition rate. Eight subjects participated in the online BCI testing, the total average accuracy recognition rate was up to 97.8%. It further verifies the potential of Chirp-VEP application in BCI control.