A sequential learning model with GNN for EEG-EMG-based stroke rehabilitation BCI

Haoyang Li; Hongfei Ji; Jian Yu; Jie Li; Lingjing Jin; Lingyu Liu; Zhongfei Bai; Chen Ye

doi:10.3389/fnins.2023.1125230

A sequential learning model with GNN for EEG-EMG-based stroke rehabilitation BCI

Front Neurosci. 2023 Apr 17:17:1125230. doi: 10.3389/fnins.2023.1125230. eCollection 2023.

Authors

Haoyang Li¹, Hongfei Ji¹, Jian Yu¹, Jie Li¹, Lingjing Jin^{2

3}, Lingyu Liu², Zhongfei Bai², Chen Ye¹

Affiliations

¹ Translational Research Center, Shanghai Yangzhi Rehabilitation Hospital (Shanghai Sunshine Rehabilitation Center), School of Electronic and Information Engineering, Tongji University, Shanghai, China.
² Department of Neurology and Neurological Rehabilitation, Shanghai Disabled Person's Federation Key Laboratory of Intelligent Rehabilitation Assistive Devices and Technologies, Yangzhi Rehabilitation Hospital (Shanghai Sunshine Rehabilitation Center), School of Medicine, Tongji University, Shanghai, China.
³ Neurotoxin Research Center of Key Laboratory of Spine and Spinal Cord Injury Repair and Regeneration of Ministry of Education, Neurological Department of Tongji Hospital, School of Medicine, Tongji University, Shanghai, China.

Abstract

Introduction: Brain-computer interfaces (BCIs) have the potential in providing neurofeedback for stroke patients to improve motor rehabilitation. However, current BCIs often only detect general motor intentions and lack the precise information needed for complex movement execution, mainly due to insufficient movement execution features in EEG signals.

Methods: This paper presents a sequential learning model incorporating a Graph Isomorphic Network (GIN) that processes a sequence of graph-structured data derived from EEG and EMG signals. Movement data are divided into sub-actions and predicted separately by the model, generating a sequential motor encoding that reflects the sequential features of the movements. Through time-based ensemble learning, the proposed method achieves more accurate prediction results and execution quality scores for each movement.

Results: A classification accuracy of 88.89% is achieved on an EEG-EMG synchronized dataset for push and pull movements, significantly outperforming the benchmark method's performance of 73.23%.

Discussion: This approach can be used to develop a hybrid EEG-EMG brain-computer interface to provide patients with more accurate neural feedback to aid their recovery.

Keywords: brain-computer interfaces; corticomuscular coherence; graph neural network; neurofeedback; sequential learning; stroke rehabilitation.

Grants and funding

This work was supported by the Shanghai Municipal Science and Technology Major Project (2021SHZDZX0100), the Fundamental Research Funds for the Central Universities, the Science and Technology Innovation Action Plan of the Shanghai Science and Technology Commission (19441908000), and Program of Shanghai Academic Research Leader (20XD1403400).