传感技术学报2025，Vol.38Issue(11)：1978-1989,12.DOI:10.3969/j.issn.1004-1699.2025.11.008
基于同步性及节点特性的通道选择对MI-EEG分类的影响研究
Research on the Influence of Channel Selection Based on Synchronization and Node Properties in MI-EEG Classification
摘要
Abstract
To effectively extract motor imagery electroencephalography(MI-EEG)features and retain sufficient information,and seek a way to improve the classification accuracy of deep learning for MI-EEG,a channel selection method based on synchronization and node proper-ties by considering the density of channel information is proposed.The phase locked value(PLV)is introduced to construct the brain net-work,and complete the classification and screening of each node in the brain network based on correlation.According to the density of channel information,20 channels from 32 EEG channels related to MI research are selected as the inputs of the convolutional neural net-work(CNN)for the comparison of offline performance classification.After studying the effect of screening channels on MI-EEG classifica-tion,it is found that the average classification accuracy of MI-EEG is enhanced from 86.29%to 90.46%for 10 recruited subjects,and this index ranges from 85.30%to 93.56%for 4 samples from public data sets.Results analysis from the change of EEG energy and brain net-works collectivization perspective reveals the channel selection influence on the properties of CNN classification for MI-EEG.关键词
运动想象脑电/脑网络/通道选择/锁相值/卷积神经网络Key words
motor imagery electroencephalography/brain network/channel selection/phase locking value/convolutional neural network分类
信息技术与安全科学引用本文复制引用
郗凯,凌鹏,陈鹏,李思敏..基于同步性及节点特性的通道选择对MI-EEG分类的影响研究[J].传感技术学报,2025,38(11):1978-1989,12.基金项目
四川省重点研发计划项目(2022YFS0021,2023YFH0037) （2022YFS0021,2023YFH0037）
