一种高效运动想象脑电信号浅层卷积解码网络  

An Efficient Shallow Convolutional Decoding Network for Motor Imagery Electroencephalography Signals

作　　者：李文平 徐光华[1,2] 张凯[1] 张四聪[1] 赵丽娇 李辉[1] LI Wenping;XU Guanghua;ZHANG Kai;ZHANG Sicong;ZHAO Lijiao;LI Hui(School of Mechanical Engineering,Xi’an Jiaotong University,Xi’an 710049,China;State Key Laboratory of Mechanical Manufacturing System Engineering,Xi’an Jiaotong University,Xi’an 710054,China;School of Economics and Finance,Xi’an Jiaotong University,Xi’an 710061,China)

机构地区：[1]西安交通大学机械工程学院,西安710049 [2]西安交通大学机械制造系统工程国家重点实验室,西安710054 [3]西安交通大学经济与金融学院,西安710061

出　　处：《西安交通大学学报》2023年第10期11-19,共9页Journal of Xi'an Jiaotong University

基　　金：国家重点研发计划资助项目(2021ZD0204300);陕西省重点研发计划资助项目(2021GXLH-Z-008);广州市科技计划资助项目(202206060003)。

摘　　要：针对现有运动想象脑机接口(MI-BCIs)中,基于深度学习的脑电信号解码网络(EEGNet)时域-空域-频域耦合特征学习能力差、模型训练与推理时间长的问题,提出了一种高效运动想象脑电信号浅层卷积解码网络(Faster-EEGNet)。该网络将第1层二维平面串行卷积优化为所有通道同时进行的串行卷积,完成了各通道信号的时域滤波与空间滤波;在中间深度卷积层对空间模式提取信号进行时域卷积特征提取,然后由深度分离卷积再次提取信号的时间-空间耦合特征,并对其进行模式识别。采用公开数据集进行仿真实验验证,结果表明:Faster-EEGNet网络的运动想象识别准确率与信息传输率相较于EEGNet网络有更好的表现,在本实验的小样本训练场景下也能够取得较好的识别效果;相较于EEGNet网络,Faster-EEGNet网络的训练时间减少了44.8%,模型推理时间减少了43.6%以上。实验结果证明所提Faster-EEGNet网络能够提升运动想象脑机接口系统的识别准确性、便捷性及快速响应性能。To address the problems of poor temporal-spatial-frequency coupling feature learning and lengthy model training and inference of deep learning-based electroencephalography signal decoding networks(EEGNet)in existing motor imagery-based brain-computer interfaces(MI-BCIs),an efficient shallow convolutional decoding network called Faster-EEGNet was proposed.In this network,the first layer of two-dimensional serial convolution was optimized to perform parallel convolution on all channels simultaneously.This enabled the completion of temporal filtering and spatial filtering of signals across all channels.Temporal convolutional features were captured from spatial pattern-extracted signals at the intermediate deep convolutional layers.Then,the depthwise separable convolution was used to capture the temporal-spatial coupling features of the signals for pattern recognition.Experimental validation was conducted using publicly available datasets.The results demonstrate that the Faster-EEGNet exhibits better performance than the EEGNet in motor imagery recognition accuracy and information transfer rate.This network also achieves good recognition results in small-sample training scenarios.Furthermore,the Faster-EEGNet reduces training time by 44.8%and model inference time by more than 43.6%compared to the EEGNet.These findings demonstrate that the proposed Faster-EEGNet can enhance the recognition accuracy,convenience,and rapid response performance of the motor imagery brain-computer interface system.

关 键 词：脑电信号 脑机接口 运动想象 深度学习 脑电解码算法 

分 类 号：Q983.5[生物学—人类学] TP242[自动化与计算机技术—检测技术与自动化装置]