Schudlo Larissa C, Chau Tom
a Bloorview Research Institute , Holland Bloorview Kids Rehabilitation Hospital , Toronto , Canada.
b Institute of Biomaterials and Biomedical Engineering , University of Toronto , Toronto , Canada.
Disabil Rehabil Assist Technol. 2018 Aug;13(6):581-591. doi: 10.1080/17483107.2017.1357212. Epub 2017 Jul 31.
For non-verbal individuals, brain-computer interfaces (BCIs) are a potential means of communication. Near-infrared spectroscopy (NIRS) is a brain-monitoring modality that has been considered for BCIs. To date, limited NIRS-BCI testing has involved online classification, particularly with individuals with severe motor impairments.
We tested an online NIRS-BCI developed for a non-verbal individual with severe congenital motor impairments. The binary BCI differentiated categorical verbal fluency task (VFT) performance and rest using prefrontal measurements. The participant attended five sessions, the last two of which were online with classification feedback.
An online classification accuracy of 63.33% was achieved using a linear discriminant classifier trained on a four-dimensional feature set. An offline, cross-validation analysis of all data yielded an optimal adjusted classification accuracy of 66.6 ± 9.11%. Inconsistent functional responses, contradictory effects of feedback, participant fatigue and motion artefacts were identified as challenges to online classification specific to this participant.
Results suggest potential in using an NIRS-BCI controlled by the VFT in instances of severe congenital impairments. Further testing with users with severe disabilities is necessary. Implications for Rehabilitation Brain-computer interfaces (BCIs) can provide a non-motor based means of communication for individuals with severe motor impairments. Near-infrared spectroscopy (NIRS) is a haemodynamic-based brain-imaging modality used in BCIs. To date, NIRS-BCIs have not been thoroughly tested with potential target users. This case study shows that NIRS-BCIs may offer a means of practical communication for individuals with severe congenital impairments and continued exploration is advisable.
对于无语言能力的个体，脑机接口（BCI）是一种潜在的交流方式。近红外光谱（NIRS）是一种已被考虑用于脑机接口的脑监测方式。迄今为止，有限的NIRS-BCI测试涉及在线分类，特别是针对有严重运动障碍的个体。
我们测试了一种为一名患有严重先天性运动障碍的无语言能力个体开发的在线NIRS-BCI。二元脑机接口利用前额叶测量来区分分类语言流畅性任务（VFT）表现和静息状态。参与者参加了五次实验，其中最后两次是有分类反馈的在线实验。
使用在四维特征集上训练的线性判别分类器，实现了63.33%的在线分类准确率。对所有数据进行的离线交叉验证分析得出的最佳调整分类准确率为66.6±9.11%。功能反应不一致、反馈的矛盾影响、参与者疲劳和运动伪影被确定为该参与者在线分类特有的挑战。
结果表明，在严重先天性损伤的情况下，使用由语言流畅性任务控制的NIRS-BCI具有潜力。有必要对重度残疾用户进行进一步测试。康复意义脑机接口（BCI）可为严重运动障碍个体提供一种基于非运动的交流方式。近红外光谱（NIRS）是一种用于脑机接口的基于血液动力学的脑成像方式。迄今为止，NIRS-BCI尚未在潜在目标用户中进行全面测试。本案例研究表明，NIRS-BCI可能为严重先天性损伤个体提供一种实际的交流方式，建议继续进行探索。
