We observed the steady-state visually evoked potential(SSVEP) from a healthy subject using a compact quad-channel potassium spin exchange relaxation-free(SERF) optically pumped magnetometer(OPM). To this end, 30 s of ...We observed the steady-state visually evoked potential(SSVEP) from a healthy subject using a compact quad-channel potassium spin exchange relaxation-free(SERF) optically pumped magnetometer(OPM). To this end, 30 s of data were collected, and SSVEP-related magnetic responses with signal intensity ranging from 150 fT to 300 f T were observed for all four channels. The corresponding signal to noise ratio(SNR) was in the range of 3.5–5.5. We then used different channels to operate the sensor as a gradiometer. In the specific case of detecting SSVEP, we noticed that the short channel separation distance led to a strongly diminished gradiometer signal. Although not optimal for the case of SSVEP detection, this set-up can prove to be highly useful for other magnetoencephalography(MEG) paradigms that require good noise cancellation.Considering its compactness, low cost, and good performance, the K-SERF sensor has great potential for biomagnetic field measurements and brain-computer interfaces(BCI).展开更多
针对卷积神经网络(CNN)在感受野有限、缺乏对全局信息的有效感知,以及在处理短时稳态运动视觉诱发电位(SSMVEP)信号时分类效果欠佳的问题,提出了一种紧凑EEGNet-Transformer(即EEGNetformer)网络。EEGNetformer网络融合了为脑电(EEG)信...针对卷积神经网络(CNN)在感受野有限、缺乏对全局信息的有效感知,以及在处理短时稳态运动视觉诱发电位(SSMVEP)信号时分类效果欠佳的问题,提出了一种紧凑EEGNet-Transformer(即EEGNetformer)网络。EEGNetformer网络融合了为脑电(EEG)信号识别任务而设计的通用的卷积神经网络EEGNet网络和Transformer网络的优势,有效地捕捉与处理脑电信号中的局部和全局信息,增强网络对SSMVEP特征的学习,进而实现良好的解码性能。EEGNet网络用于提取SSMVEP的局部时间和空间特征,而Transformer网络用于捕捉脑电时间序列的全局信息。在基于SSMVEP-BCI范式采集的数据基础上,开展了实验以评估EEGNetformer网络的性能。实验结果显示,当在2 s SSMVEP数据条件下,EEGNetformer网络在基于被试者内情况的平均准确率为88.9%±6.6%,在基于跨被试者情况的平均准确率为69.1%±4.3%。与传统的CNN算法相比,EEGNetformer网络的分类性能提升了4.2%~17.4%。研究内容说明,EEGNetformer网络在有效提高SSMVEP-BCI识别准确率方面具有显著优势,为进一步提升SSMVEP-BCI解码性能提供了新的研究思路。展开更多
-Brain-computer interface （BCI） can help the deformity person finish some basic activities. In this paper, we concern some critical aspects of SSVEP based BCI, including stimulator selection, method of SSVEP extract...-Brain-computer interface （BCI） can help the deformity person finish some basic activities. In this paper, we concern some critical aspects of SSVEP based BCI, including stimulator selection, method of SSVEP extracting in a short time, stimulating frequency selection, and signal electrode selection. The conclusion is that the stimulator type should be based on the complexity of the BCI system, the method based on wavelet analysis is more valid than the power spectrum method in extracting the SSVEP in a short period, and the selections of stimulating frequency and electrode are important in designing a BCI system. These contents are meaningful for implementing a real SSVEP-based BCI.展开更多
针对稳态视觉诱发电位(steady state visual evoked potential,SSVEP)范式下脑电信号(electroencephalograph,EEG)信噪比低、限制其识别正确率提高及脑-机接口应用等问题,根据EEG随机性、近似平稳的特点,提出了用于SSVEP特征频率提取的...针对稳态视觉诱发电位(steady state visual evoked potential,SSVEP)范式下脑电信号(electroencephalograph,EEG)信噪比低、限制其识别正确率提高及脑-机接口应用等问题,根据EEG随机性、近似平稳的特点,提出了用于SSVEP特征频率提取的同步压缩短时傅里叶变换方法。该方法利用短时傅里叶变换对EEG进行时频分析,并通过同步压缩变换对时频平面的能量在频率方向进行重新分配,获得频率曲线更加集中的时频表达;同时为提高EEG信噪比,提取SSVEP脑电中特征频率附近信号进行重构,并利用典型相关分析进行分类识别,有效提高了最终识别正确率。仿真和实验结果表明,该方法极大地提高了信号的信噪比,具有良好的抗噪声性能和信号提取精度,且与传统的经验模态分解和常规滤波方法相比,该方法平均识别正确率最多分别提高了9.98%和4.38%,平均信息传输率最多分别提高了7.57bit/min和2.69bit/min,有效提高了SSVEP范式下脑-机接口的工作性能。展开更多
基金Project supported by the National Key Research and Development Program of China(Grant Nos.2016YFA0300600 and 2016YFA0301500)the Strategic Priority Research Program of Chinese Academy of Sciences(Grant Nos.XDB07030000 and XDBS32000000)+1 种基金the National Natural Science Foundation of China(Grant Nos.11474347 and 31730039)the Fund from the Ministry of Science and Technology of China(Grant No.2015CB351701)
文摘We observed the steady-state visually evoked potential(SSVEP) from a healthy subject using a compact quad-channel potassium spin exchange relaxation-free(SERF) optically pumped magnetometer(OPM). To this end, 30 s of data were collected, and SSVEP-related magnetic responses with signal intensity ranging from 150 fT to 300 f T were observed for all four channels. The corresponding signal to noise ratio(SNR) was in the range of 3.5–5.5. We then used different channels to operate the sensor as a gradiometer. In the specific case of detecting SSVEP, we noticed that the short channel separation distance led to a strongly diminished gradiometer signal. Although not optimal for the case of SSVEP detection, this set-up can prove to be highly useful for other magnetoencephalography(MEG) paradigms that require good noise cancellation.Considering its compactness, low cost, and good performance, the K-SERF sensor has great potential for biomagnetic field measurements and brain-computer interfaces(BCI).
文摘针对卷积神经网络(CNN)在感受野有限、缺乏对全局信息的有效感知,以及在处理短时稳态运动视觉诱发电位(SSMVEP)信号时分类效果欠佳的问题,提出了一种紧凑EEGNet-Transformer(即EEGNetformer)网络。EEGNetformer网络融合了为脑电(EEG)信号识别任务而设计的通用的卷积神经网络EEGNet网络和Transformer网络的优势,有效地捕捉与处理脑电信号中的局部和全局信息,增强网络对SSMVEP特征的学习,进而实现良好的解码性能。EEGNet网络用于提取SSMVEP的局部时间和空间特征,而Transformer网络用于捕捉脑电时间序列的全局信息。在基于SSMVEP-BCI范式采集的数据基础上,开展了实验以评估EEGNetformer网络的性能。实验结果显示,当在2 s SSMVEP数据条件下,EEGNetformer网络在基于被试者内情况的平均准确率为88.9%±6.6%,在基于跨被试者情况的平均准确率为69.1%±4.3%。与传统的CNN算法相比,EEGNetformer网络的分类性能提升了4.2%~17.4%。研究内容说明,EEGNetformer网络在有效提高SSMVEP-BCI识别准确率方面具有显著优势,为进一步提升SSMVEP-BCI解码性能提供了新的研究思路。
基金supported by the National Natural Science Foundation of China under Grant No. 30525030 and60736029.
文摘-Brain-computer interface （BCI） can help the deformity person finish some basic activities. In this paper, we concern some critical aspects of SSVEP based BCI, including stimulator selection, method of SSVEP extracting in a short time, stimulating frequency selection, and signal electrode selection. The conclusion is that the stimulator type should be based on the complexity of the BCI system, the method based on wavelet analysis is more valid than the power spectrum method in extracting the SSVEP in a short period, and the selections of stimulating frequency and electrode are important in designing a BCI system. These contents are meaningful for implementing a real SSVEP-based BCI.
文摘针对稳态视觉诱发电位(steady state visual evoked potential,SSVEP)范式下脑电信号(electroencephalograph,EEG)信噪比低、限制其识别正确率提高及脑-机接口应用等问题,根据EEG随机性、近似平稳的特点,提出了用于SSVEP特征频率提取的同步压缩短时傅里叶变换方法。该方法利用短时傅里叶变换对EEG进行时频分析,并通过同步压缩变换对时频平面的能量在频率方向进行重新分配,获得频率曲线更加集中的时频表达;同时为提高EEG信噪比,提取SSVEP脑电中特征频率附近信号进行重构,并利用典型相关分析进行分类识别,有效提高了最终识别正确率。仿真和实验结果表明,该方法极大地提高了信号的信噪比,具有良好的抗噪声性能和信号提取精度,且与传统的经验模态分解和常规滤波方法相比,该方法平均识别正确率最多分别提高了9.98%和4.38%,平均信息传输率最多分别提高了7.57bit/min和2.69bit/min,有效提高了SSVEP范式下脑-机接口的工作性能。
