Department of Medical IT Convergence Engineering, Kumoh National Institute of Technology, Daehak-ro 61, Gumi 39177, Republic of Korea.
Machine Learning Group, Berlin Institute of Technology (TU Berlin), Marchstrasse 23, Berlin 10587, Germany.
Gigascience. 2019 Nov 1;8(11). doi: 10.1093/gigascience/giz133.
A steady-state visual-evoked potential (SSVEP) is a brain response to visual stimuli modulated at certain frequencies; it has been widely used in electroencephalography (EEG)-based brain-computer interface research. However, there are few published SSVEP datasets for brain-computer interface. In this study, we obtained a new SSVEP dataset based on measurements from 30 participants, performed on 2 days; our dataset complements existing SSVEP datasets: (i) multi-band SSVEP datasets are provided, and all 3 possible frequency bands (low, middle, and high) were used for SSVEP stimulation; (ii) multi-day datasets are included; and (iii) the EEG datasets include simultaneously obtained physiological measurements, such as respiration, electrocardiography, electromyography, and head motion (accelerator).
To validate our dataset, we estimated the spectral powers and classification performance for the EEG (SSVEP) datasets and created an example plot to visualize the physiological time-series data. Strong SSVEP responses were observed at stimulation frequencies, and the mean classification performance of the middle frequency band was significantly higher than the low- and high-frequency bands. Other physiological data also showed reasonable results.
Our multi-band, multi-day SSVEP datasets can be used to optimize stimulation frequencies because they enable simultaneous investigation of the characteristics of the SSVEPs evoked in each of the 3 frequency bands, and solve session-to-session (day-to-day) transfer problems by enabling investigation of the non-stationarity of SSVEPs measured on different days. Additionally, auxiliary physiological data can be used to explore the relationship between SSVEP characteristics and physiological conditions, providing useful information for optimizing experimental paradigms to achieve high performance.
稳态视觉诱发电位（SSVEP）是大脑对调制特定频率的视觉刺激的反应，它已被广泛应用于基于脑电图（EEG）的脑机接口研究。然而，用于脑机接口的 SSVEP 数据集很少。在这项研究中，我们基于 30 名参与者的测量结果获得了一个新的 SSVEP 数据集，这些数据是在两天内采集的；我们的数据集补充了现有的 SSVEP 数据集：（i）提供多频段 SSVEP 数据集，所有 3 个可能的频段（低、中、高）都用于 SSVEP 刺激；（ii）包含多日数据集；（iii）EEG 数据集包括同时获得的生理测量，如呼吸、心电图、肌电图和头部运动（加速度计）。
为了验证我们的数据集，我们对 EEG（SSVEP）数据集进行了频谱功率估计和分类性能评估，并创建了一个示例图来可视化生理时间序列数据。在刺激频率处观察到强烈的 SSVEP 反应，中频带的平均分类性能明显高于低频带和高频带。其他生理数据也显示出合理的结果。
我们的多频段、多日 SSVEP 数据集可用于优化刺激频率，因为它们可以同时研究在这 3 个频段中每个频段诱发的 SSVEP 的特征，并通过研究不同天测量的 SSVEP 的非平稳性来解决会话间（每天之间）的转移问题。此外，辅助生理数据可用于探索 SSVEP 特征与生理状况之间的关系，为优化实验范式以实现高性能提供有用信息。
