The study developed a memory task training system using functional near-infrared spectroscopy (fNIRS) and neurofeedback mechanisms, and acquired and analyzed subjects' EEG signals. The results showed that subjects participating in the neurofeedback task had higher correlated brain network node degrees and average cluster coefficients in the right hemisphere brain region of the prefrontal lobe, with relatively lower dispersion of mediator centrality. In addition, the subjects' left hemisphere brain region of the prefrontal lobe section had increased centrality in the neurofeedback task. Classification of brain data by the channel network model and the support vector machine model showed that the classification accuracy of both models was higher in the task state and resting state than in the feedback task and the control task, and the classification accuracy of the channel network model was higher. The results suggested that subjects in the neurofeedback task had distinct brain data features and that these features could be effectively recognized.
该研究使用功能性近红外光谱技术和神经反馈机制开发了记忆任务训练系统，并对被试的脑电信号进行采集和分析。结果显示，参与神经反馈任务的被试在前额叶部右半脑区具有更高的相关性脑网络节点度和平均集群系数，中介中心度的分散程度相对降低。此外，被试的前额叶部左半脑区在神经反馈任务中的核心地位上升。通过通道网络模型和支持向量机模型对脑数据进行分类，结果表明两个模型在任务态和静息态的分类准确率高于反馈任务和对照任务，且通道网络模型的分类准确率更高。该研究结果表明神经反馈任务的被试具有明显的脑数据特征，并且这些特征能够被有效识别。.
Keywords: brain network; channel network model; functional near-infrared spectroscopy; neurofeedback.