Weyand Sabine, Takehara-Nishiuchi Kaori, Chau Tom
Bloorview Research Institute, Holland Bloorview Kids Rehabilitation Hospital, 150 Kilgour Road, Toronto, ON, Canada M4G 1R8; Institute of Biomaterials and Biomedical Engineering, University of Toronto, 164 College St, Toronto, ON, Canada M5S 3G9.
Department of Psychology, University of Toronto, 100 St. George Street, Toronto, ON, Canada M5S 3G3.
J Neurosci Methods. 2015 Oct 30;254:36-45. doi: 10.1016/j.jneumeth.2015.07.007. Epub 2015 Jul 20.
Near-infrared spectroscopy (NIRS) brain-computer interfaces (BCIs) enable users to interact with their environment using only cognitive activities. This paper presents the results of a comparison of four methodological frameworks used to select a pair of tasks to control a binary NIRS-BCI; specifically, three novel personalized task paradigms and the state-of-the-art prescribed task framework were explored.
Three types of personalized task selection approaches were compared, including: user-selected mental tasks using weighted slope scores (WS-scores), user-selected mental tasks using pair-wise accuracy rankings (PWAR), and researcher-selected mental tasks using PWAR. These paradigms, along with the state-of-the-art prescribed mental task framework, where mental tasks are selected based on the most commonly used tasks in literature, were tested by ten able-bodied participants who took part in five NIRS-BCI sessions.
The frameworks were compared in terms of their accuracy, perceived ease-of-use, computational time, user preference, and length of training. Most notably, researcher-selected personalized tasks resulted in significantly higher accuracies, while user-selected personalized tasks resulted in significantly higher perceived ease-of-use. It was also concluded that PWAR minimized the amount of data that needed to be collected; while, WS-scores maximized user satisfaction and minimized computational time.
In comparison to the state-of-the-art prescribed mental tasks, our findings show that overall, personalized tasks appear to be superior to prescribed tasks with respect to accuracy and perceived ease-of-use.
The deployment of personalized rather than prescribed mental tasks ought to be considered and further investigated in future NIRS-BCI studies.
近红外光谱（NIRS）脑机接口（BCI）使用户能够仅通过认知活动与周围环境进行交互。本文展示了用于选择一对任务以控制二进制NIRS-BCI的四种方法框架的比较结果；具体而言，探索了三种新颖的个性化任务范式和最先进的规定任务框架。
比较了三种类型的个性化任务选择方法，包括：使用加权斜率分数（WS分数）的用户选择心理任务、使用成对准确率排名（PWAR）的用户选择心理任务以及使用PWAR的研究人员选择心理任务。这些范式与最先进的规定心理任务框架（其中心理任务是根据文献中最常用的任务选择的）一起，由10名身体健全的参与者进行测试，他们参加了五次NIRS-BCI实验。
从准确率、感知易用性、计算时间、用户偏好和训练时长方面对这些框架进行了比较。最值得注意 的是，研究人员选择的个性化任务准确率显著更高，而用户选择的个性化任务感知易用性显著更高。还得出结论，PWAR将需要收集的数据量减到最少；而WS分数使用户满意度最大化并使计算时间最小化。
与最先进的规定心理任务相比，我们的研究结果表明，总体而言，个性化任务在准确率和感知易用性方面似乎优于规定任务。
在未来的NIRS-BCI研究中，应考虑并进一步研究采用个性化而非规定性心理任务。
Zotero 插件