Luo Xi, Lin Yanfei, Guo Rongxiao, Gao Xiaorong, Zhang Shangen
IEEE Trans Neural Syst Rehabil Eng. 2023;31:1933-1942. doi: 10.1109/TNSRE.2023.3263502. Epub 2023 Apr 5.
Hybrid brain-computer interfaces (HBCI) combining eye-tracker has attracted the attentions of researchers in target recognition. However, there are still many issues to be addressed in rapid sequence visual presentation (RSVP) tasks, such as the effect of presentation rates and target types on event-related potentials (ERP) and pupillometry, synchronization analysis of electroencephalography (EEG) and eye-tracking, and so on. In this study, the RSVP experiments with three different target types of pictures, words and numbers at the presentation rates of 100 and 200 ms were conducted. EEG data and pupillometry data were synchronously collected from 20 university students. The results of ERP analysis showed that, among three different target types at the presentation rate of 100 ms, the picture P300 component had the largest amplitude and the longest latency. From the 100 ms presentation rates to 200 ms one for the three target types, the P300 amplitudes became smaller, and the P300 latencies became shorter. The results of pupillometry analysis showed that, at the presentation rates of 100 and 200 ms, the pupil dilation of pictures had the smallest amplitude and the shortest latency. At the two presentation rates, no significant differences of pupil size and latency were found for the three target types. For the early pupil dilation within 1000 ms, the picture pupil size was significantly smaller than the other ones, and the picture pupil acceleration had the largest average amplitude and the shortest latency. These pupillometry features within 1000 ms combining with the P300 features could be taken as the effective ones for target classification. Through synchronization analysis of the EEG data and pupillometry data, the effects of target type and presentation rate on ERP and pupil dilation were different. These results could contribute to developing the fusion methods between EEG and eye-tracking, and provide valuable references for the multi-target recognition of hybrid BCI based on eye-tracking.
结合眼动追踪器的混合脑机接口（HBCI）在目标识别方面引起了研究人员的关注。然而，在快速序列视觉呈现（RSVP）任务中仍有许多问题有待解决，例如呈现速率和目标类型对事件相关电位（ERP）和瞳孔测量的影响、脑电图（EEG）与眼动追踪的同步分析等。在本研究中，进行了以100毫秒和200毫秒的呈现速率呈现三种不同目标类型（图片、单词和数字）的RSVP实验。从20名大学生中同步收集了EEG数据和瞳孔测量数据。ERP分析结果表明，在100毫秒的呈现速率下，三种不同目标类型中，图片的P300成分振幅最大、潜伏期最长。对于三种目标类型，从100毫秒的呈现速率到200毫秒的呈现速率，P300振幅变小，P300潜伏期变短。瞳孔测量分析结果表明，在100毫秒和200毫秒的呈现速率下，图片的瞳孔扩张振幅最小、潜伏期最短。在这两种呈现速率下，三种目标类型的瞳孔大小和潜伏期均未发现显著差异。对于1000毫秒内的早期瞳孔扩张，图片的瞳孔大小明显小于其他类型，且图片的瞳孔加速度平均振幅最大、潜伏期最短。1000毫秒内的这些瞳孔测量特征与P300特征可作为目标分类的有效特征。通过对EEG数据和瞳孔测量数据的同步分析，目标类型和呈现速率对ERP和瞳孔扩张的影响不同。这些结果有助于开发EEG与眼动追踪之间的融合方法，并为基于眼动追踪的混合BCI多目标识别提供有价值的参考。
