Decoding human visual colour EEG information using machine learning and visual evoked potentials

With the rapid development of brain-computer interfaces (BCIs), human visual decoding, one of the important research directions of BCIs, has attracted a substantial amount of attention. However, most visual decoding studies have focused on graphic and image decoding. In this paper, we first demonstrate the possibility of building a new kind of task-irrelevant, simple and fast-stimulus BCI-based experimental paradigm that relies on visual evoked potentials (VEPs) during colour observation. Additionally, the features of visual colour information were found through reliable real-time decoding. We selected 9 subjects who did not have colour blindness to participate in our tests. These subjects were asked to observe red, green, and blue screens in turn with an interstimulus interval of 1 second. The machine learning results showed that the visual colour classification accuracy had a maximum of 93.73%. The latency evoked by visual colour stimuli was within the P300 range, i.e., 176.8 milliseconds for the red screen, 206.5 milliseconds for the green screen, and 225.3 milliseconds for the blue screen. The experimental results hereby show that the VEPs can be used for reliable colour real-time decoding.

随着脑机接口（Brain-Computer Interface, BCI）技术的快速发展，作为脑机接口重要研究方向之一的人类视觉解码，已受到学界广泛关注。然而，当前多数视觉解码研究均聚焦于图形与图像解码任务。本文首先证实了构建一类新型任务无关、简易且支持快速刺激的脑机接口实验范式的可行性，该范式依托颜色观测过程中的视觉诱发电位（Visual Evoked Potential, VEP）。此外，本研究通过可靠的实时解码，成功提取了视觉色彩信息的特征。我们招募了9名无色盲症状的受试者参与实验，要求受试者依次观测红色、绿色与蓝色屏幕，刺激间间隔为1秒。机器学习实验结果表明，视觉色彩分类的最高准确率可达93.73%。视觉色彩刺激诱发的潜伏期处于P300区间内，其中红色屏幕对应的潜伏期为176.8毫秒，绿色屏幕为206.5毫秒，蓝色屏幕为225.3毫秒。综上实验结果表明，视觉诱发电位可用于可靠的色彩实时解码任务。