fNIRS脑功能数据分析与行为分类  

fNIRS BRAIN FUNCTION DATA ANALYSIS AND BEHAVIOR CLASSIFICATION

作　　者：王景凯 周书锋[2] Wang Jingkai;Zhou Shufeng(Ji Xianlin Honors School,Liaocheng University,252000,Liaocheng,Shandong,China;School of Mathematical Science,Liaocheng University,252000,Liaocheng,Shandong,China)

机构地区：[1]聊城大学季羡林学院,山东聊城252000 [2]聊城大学数学科学学院,山东聊城252000

出　　处：《山东师范大学学报（自然科学版）》2022年第3期283-291,共9页Journal of Shandong Normal University(Natural Science)

基　　金：聊城大学大学生创新训练计划项目(CXCY2021150)。

摘　　要：随着脑功能成像技术的不断发展,功能性近红外光谱技术(fNIRS)已经被广泛应用于各个领域中.为探究针对单侧手指和脚敲击动作fNIRS信号的可分性,并相应提高分类的准确率,研究分析了30名受试者在执行三种动作(左手敲击、右手敲击和脚敲击)期间的fNIRS信号,提取了不同动作执行期间HbR和HbO信号的相关系数作为特征,构造了基于阈值筛选的脑功能网络,在此基础上运用支持向量机实施了二元和三元行为分类.此外,通过调参得到最优深度神经网络模型实施三分类,得到的最高准确率为95.65%.该研究解决了任务识别与分类的问题,有助于为fNIRS-BCI增加新的控制指令,为偏瘫患者或者肢体运动障碍者的康复训练提供一种新的选择参考.With the development of brain functional imaging technology, functional near infrared spectroscopy(fNIRS) has been widely used in various fields. In order to explore the separability of fNIRS signals for unilateral finger and foot tapping actions and to improve the accuracy of classification accordingly, the fNIRS signals of 30 subjects during the execution of three actions(LFT, RFT and FT) were analyzed. The correlation coefficients of HbR and HbO signals during the execution of different actions were extracted as features, and a brain functional network based on threshold screening was constructed. On this basis, support vector machine was used to implement binary and ternary behavior classification. In addition, the optimal depth neural network model was obtained by adjusting parameters to implement three classifications, and the highest accuracy was 95.65%. This research solves the problem of task recognition and classification, helps to add new control commands to fNIRS-BCI, and provides a new reference for rehabilitation training of hemiplegic patients or people with limb movement disorders.

关 键 词：功能性近红外光谱 数据分析 行为动作分类 深度神经网络 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]