Institute of Systems and Robotics, University of Coimbra, Coimbra, Portugal; Polytechnic Institute of Tomar, Department of Engineering, Tomar, Portugal.
Institute of Systems and Robotics, University of Coimbra, Coimbra, Portugal.
J Neurosci Methods. 2022 Sep 1;379:109661. doi: 10.1016/j.jneumeth.2022.109661. Epub 2022 Jul 8.
Brain-computer interfaces (BCIs) are a promising tool for communication with completely locked-in state (CLIS) patients. Despite the great efforts already made by the BCI research community, the cases of success are still very few, very exploratory, limited in time, and based on simple 'yes/no' paradigms.
A P300-based BCI is proposed comparing two conditions, one corresponding to purely spatial auditory stimuli (AU-S) and the other corresponding to hybrid visual and spatial auditory stimuli (HVA-S). In the HVA-S condition, there is a semantic, temporal, and spatial congruence between visual and auditory stimuli. The stimuli comprise a lexicon of 7 written and spoken words. Spatial sounds are generated through the head-related transfer function. Given the good results obtained with 10 able-bodied participants, we investigated whether a patient entering CLIS could use the proposed BCI.
The able-bodied group achieved 71.3 % and 90.5 % online classification accuracy for the auditory and hybrid BCIs respectively, while the patient achieved 30 % and chance level accuracies, for the same conditions. Notwithstanding, the patient's event-related potentials (ERPs) showed statistical discrimination between target and non-target events in different time windows.
The results of the control group compare favorably with the state-of-the-art, considering a 7-class BCI controlled visual-covertly and with auditory stimuli. The integration of visual and auditory stimuli has not been tested before with CLIS patients.
The semantic, temporal, and spatial congruence of the stimuli increased the performance of the control group, but not of the CLIS patient, which can be due to impaired attention and cognitive function. The patient's unique ERP patterns make interpretation difficult, requiring further tests/paradigms to decouple patients' responses at different levels (reflexive, perceptual, cognitive). The ERPs discrimination found indicates that a simplification of the proposed approaches may be feasible.
脑机接口（BCI）是一种与完全闭锁状态（CLIS）患者进行交流的有前途的工具。尽管 BCI 研究界已经做出了巨大的努力，但成功案例仍然很少，非常具有探索性，时间有限，并且基于简单的“是/否”范式。
提出了一种基于 P300 的 BCI，比较了两种条件，一种对应于纯粹的空间听觉刺激（AU-S），另一种对应于混合视觉和空间听觉刺激（HVA-S）。在 HVA-S 条件下，视觉和听觉刺激之间存在语义、时间和空间一致性。刺激包括一个 7 个书面和口语单词的词汇。空间声音通过头部相关传递函数产生。鉴于 10 名健康参与者获得的良好结果，我们研究了进入 CLIS 的患者是否可以使用所提出的 BCI。
健康组在线分类准确率分别为听觉和混合 BCI 的 71.3%和 90.5%，而患者在相同条件下的准确率为 30%和机会水平。尽管如此，患者的事件相关电位（ERP）在不同的时间窗口显示出对目标和非目标事件的统计区分。
考虑到使用听觉刺激的 7 类视觉隐蔽控制 BCI，控制组的结果与最新技术相比具有优势。以前从未在 CLIS 患者中测试过视觉和听觉刺激的融合。
刺激的语义、时间和空间一致性提高了对照组的性能，但对 CLIS 患者没有提高，这可能是由于注意力和认知功能受损所致。患者独特的 ERP 模式使得解释变得困难，需要进一步的测试/范式来分离患者在不同水平（反射、感知、认知）的反应。发现的 ERP 区分表明，可以简化所提出的方法。
