A novel multi-branch hybrid neural network for motor imagery EEG signal classification. (August 2022)

• Record Type: Journal Article
• Authors: Ma, Weifeng; Xue, Haojie; Sun, Xiaoyong; Mao, Sijia; Wang, Liudi; Liu, Yang; Wang, Yuchen; Lin, Xuefen
• Is Part Of: Biomedical signal processing and control. Volume 77(2022)
• Publication Date: 2022-08
• Subjects: Motor imagery -- Convolutional Neural Network (CNN) -- Bidirectional Gated Recurrent Unit (BGRU) -- Multi-branch hybrid neural network (MBHNN) -- Frequency domain segmentation swap (Seg-Swap); Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2022.103718
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: We propose a novel multi-branch hybrid neural network which is called MBHNN. Since we split the input signal into different frequency bands for processing, the EEG classification accuracy is improved compared to the single branch structure of our proposed model. In order to make full use of the rich information in the forward and backward directions, we introduce a bidirectional attention mechanism into the BGRU block of the EEG feature decoding model, which improves the efficiency of decoding. Since the EEG field lacks a large amount of training data, deep learning is a data research methodology that requires a large number of training samples, this paper proposes a novel data augmentation method for expanding EEG datasets, which is called frequency domain segmentation swap (Seg-Swap). It is worth noting that comparing the classification performance of the state-of-the-art models with the proposed model, better results are obtained on the BCI Competition IV dataset 2a and HGD, respectively. Abstract: As a typical spontaneous brain-computer interface system, motor imagery has been widely used in areas such as robot control and stroke rehabilitation. Recently, researchers have started to study and propose various Convolutional Neural Network (CNN) structures based on motor imagery signals and have obtained better decoding accuracy compared with traditional machine learning algorithms. In this paper, we focus on a four-class motor imagery task. Therefore, we propose an end-to-end novel multi-branch hybrid neural network for motor imagery EEG signal classification. Notably, we divide the input signal into four frequency bands associated with the motor imagery signal. Meanwhile, we also introduced a Bidirectional Gated Recurrent Unit (BGRU) to identify EEG features. Nevertheless, it is extremely difficult to collect high-quality EEG data and the classification accuracy is degraded due to the strict requirements of the subjects and the experimental environment. To address this issue, we propose a novel data augmentation approach for frequency domain segmentation swap (Seg-Swap) for improving EEG motor imagery signal classification accuracy. We make use of the publicly available BCI IV 2a dataset and High Gamma dataset to evaluate the decoding performance of the model. The proposed multi-branch hybrid neural network (MBHNN) achieves 86.15% and 95.04% on the both datasets. Compared with several state-of-the-art (SOA) algorithms, our proposed model has a slight improvement in classification accuracy. Experimental results show that our proposed MBHNN has higher decoding classification and stronger robustness.