Geuze Jeroen, Farquhar Jason, Desain Peter
Radboud University Nijmegen, Donders Institute for Brain, Cognition, and Behaviour, Nijmegen, The Netherlands.
PLoS One. 2014 Feb 7;9(2):e87511. doi: 10.1371/journal.pone.0087511. eCollection 2014.
This article investigates a possible Brain Computer Interface (BCI) based on semantic relations. The BCI determines which prime word a subject has in mind by presenting probe words using an intelligent algorithm. Subjects indicate when a presented probe word is related to the prime word by a single finger tap. The detection of the neural signal associated with this movement is used by the BCI to decode the prime word. The movement detector combined both the evoked (ERP) and induced (ERD) responses elicited with the movement. Single trial movement detection had an average accuracy of 67%. The decoding of the prime word had an average accuracy of 38% when using 100 probes and 150 possible targets, and 41% after applying a dynamic stopping criterium, reducing the average number of probes to 47. The article shows that the intelligent algorithm used to present the probe words has a significantly higher performance than a random selection of probes. Simulations demonstrate that the BCI also works with larger vocabulary sizes, and the performance scales logarithmically with vocabulary size.
本文研究了一种基于语义关系的脑机接口（BCI）的可能性。该脑机接口通过使用智能算法呈现探测词来确定受试者心中所想的起始词。当呈现的探测词与起始词相关时，受试者通过单指轻敲来表明。与该动作相关的神经信号的检测被脑机接口用于解码起始词。动作检测器结合了该动作引发的诱发反应（ERP）和诱导反应（ERD）。单次试验动作检测平均准确率为67%。当使用100个探测词和150个可能的目标时，起始词的解码平均准确率为38%，在应用动态停止标准后，平均探测词数量减少到47个，准确率为41%。文章表明，用于呈现探测词的智能算法的性能明显高于随机选择探测词。模拟表明，该脑机接口在更大的词汇量下也能工作，并且性能随词汇量呈对数缩放。
