Research on Neural Activation Mechanism of Motor Imagery Based on Virtual Scene

• Published in: IET Conference Proceedings
• Main Authors: Xie, Ping, Wang, Zihao, Hualei Fang, Chen, Wei, Yu, Jinxu, Fu, Zihao, He, Sifan, Chen, Xiaoling
• Format: Conference Proceeding
• Language: English
• Published: Stevenage The Institution of Engineering & Technology 18.11.2019
• Subjects: Electroencephalography; Feature extraction; Imagery; Rehabilitation; Virtual reality; Wavelet analysis
• ISBN: 9781839534430; 1839534435
• DOI: 10.1049/cp.2019.1200

Research have confirmed that specific virtual scenes in virtual rehabilitation can promote positive activation of motor imagery nerve. In order to analyze the changing trend of the activation characteristics of motor imagery neurons in different virtual scenes. In this study, six subjects were collected for 7 consecutive days of motion-imagining EEG signals in the control group (arrow scene), static (text and picture scene), dynamic (3D scene and virtual game) scenes. In subsequent analysis, the wavelet packet method was introduced to analyze the characteristics of motion regions. In this study, the α/β band was extracted to perform the power spectrum estimation and analyze the ERD/ERS characteristics. Finally, multi-level dynamic features of time, frequency and space domain were extracted and used to classify left-handed and right handed imaginary patterns by PSO-SVM model and to compare the change of recognition rate in different scenarios. The results showed that the motor cortical nerve activation area could be significantly expand by virtual scene training, and the contralateral dominant ERD in the α/β band was enhanced. In addition, the recognition of motion imagination based on dynamic virtual scenes is higher than that in static scenes. This study also provides a theoretical foundation for the virtual scene design strategy of the MI-BCI system.