Choi Yun-Joo, Kwon Oh-Sang, Kim Sung-Phil
Department of Biomedical Engineering, Ulsan National Institute of Science and Technology, Ulsan, 44919 Korea.
Cogn Neurodyn. 2023 Dec;17(6):1401-1416. doi: 10.1007/s11571-022-09901-3. Epub 2022 Nov 18.
Non-invasive brain-computer interfaces (BCIs) based on an event-related potential (ERP) component, P300, elicited via the oddball paradigm, have been extensively developed to enable device control and communication. While most P300-based BCIs employ visual stimuli in the oddball paradigm, auditory P300-based BCIs also need to be developed for users with unreliable gaze control or limited visual processing. Specifically, auditory BCIs without additional visual support or multi-channel sound sources can broaden the application areas of BCIs. This study aimed to design optimal stimuli for auditory BCIs among artificial (e.g., beep) and natural (e.g., human voice and animal sounds) sounds in such circumstances. In addition, it aimed to investigate differences between auditory and visual stimulations for online P300-based BCIs. As a result, natural sounds led to both higher online BCI performance and larger differences in ERP amplitudes between the target and non-target compared to artificial sounds. However, no single type of sound offered the best performance for all subjects; rather, each subject indicated different preferences between the human voice and animal sound. In line with previous reports, visual stimuli yielded higher BCI performance (average 77.56%) than auditory counterparts (average 54.67%). In addition, spatiotemporal patterns of the differences in ERP amplitudes between target and non-target were more dynamic with visual stimuli than with auditory stimuli. The results suggest that selecting a natural auditory stimulus optimal for individual users as well as making differences in ERP amplitudes between target and non-target stimuli more dynamic may further improve auditory P300-based BCIs.
The online version contains supplementary material available at 10.1007/s11571-022-09901-3.
基于事件相关电位（ERP）成分P300、通过Oddball范式诱发的非侵入性脑机接口（BCI）已得到广泛开发，以实现设备控制和通信。虽然大多数基于P300的BCI在Oddball范式中采用视觉刺激，但对于注视控制不可靠或视觉处理能力有限的用户，也需要开发基于听觉P300的BCI。具体而言，无需额外视觉支持或多通道声源的听觉BCI可以拓宽BCI的应用领域。本研究旨在在此类情况下，在人工声音（如蜂鸣声）和自然声音（如人声和动物声音）中为听觉BCI设计最佳刺激。此外，它旨在研究基于在线P300的BCI中听觉和视觉刺激之间的差异。结果表明，与人工声音相比，自然声音导致更高的在线BCI性能以及目标与非目标之间ERP振幅的更大差异。然而，没有单一类型的声音对所有受试者都表现出最佳性能；相反，每个受试者在人声和动物声音之间表现出不同的偏好。与先前的报告一致，视觉刺激产生的BCI性能（平均77.56%）高于听觉刺激（平均54.67%）。此外，目标与非目标之间ERP振幅差异的时空模式在视觉刺激下比在听觉刺激下更具动态性。结果表明，选择适合个体用户的自然听觉刺激以及使目标与非目标刺激之间的ERP振幅差异更具动态性，可能会进一步改善基于听觉P300的BCI。
在线版本包含可在10.1007/s11571-022-09901-3获取的补充材料。
