基于可穿戴式脑磁图的视觉混合脑机接口研究

Study of Visual Hybrid Brain-Computer Interface Based on Wearable Magnetoencephalogram

新兴的可穿戴式脑磁图技术为脑机接口(BCI)提供高质量数据奠定了基础.为探究可穿戴式脑磁图应用于视觉混合BCI上的可行性,本文基于稳态视觉诱发场(SSVEF)和Alpha波设计了SSVEF-Alpha混合BCI,并在不同分类模型上进行了性能对比结果表明,基于用户依赖(UD)的训练方法,混合BCI6分类平均分类准确率为(93.29±1.69)%,信息传输速率可达86.81bits/min,且使用短数据长度进行用户独立(UI)的训练方法比免训练的方法更具优越性.本研究验证了视觉混合BCI的有效性,为进一步开发设计可穿戴式脑磁图的BCI应用产品提供参考范例.

The emerging wearable magnetoencephalography technology lays the foundation for brain-computer interface to provide high-quality data.To explore the feasibility of applying wearable magnetoencephalography in visual hybrid brain-computer interface,a SSVEF-Alpha hybrid brain-computer interface is designed based on steady-state visual evoked field and Alpha wave,and the performance is compared with different classification models.The results show that based on the user-dependent training method,the average classification accuracy of hybrid brain-computer interface is(93.29±1.69)%,the information transmission rate can reach 86.81 bits/min.And the user-independent training method with short data length shows superiority over the training-free method.This study verifies the effectiveness of visual hybrid brain-computer interface and provides a reference example for further development and design of brain-computer interface products of wearable magnetoencephalography.