Eeg Feature Recognition And Brain Network Mechanism Based On Steady-state Somatosensory Evoked Potential

The Brain Computer Interface(BCI)is a brain-independent normal output pathway that directly extracts information from the brain to interact with the outside world,thereby improving the patient's quality of life.In the current technology,the universality and classification accuracy of the BCI paradigm are still the main problems that hinder its entry into practical applications.Commonly used motor imaging(Motor Imagery,MI),Steady-State Visual Evoked Potential(SSVEP)and P300 paradigm BCI can not solve these two problems at the same time.This topic will study the brain-computer interface based on Steady-State Somatosensory Evoked Potential(SSSEP),and optimize the experimental parameters,improve feature extraction and feature recognition methods,and presents a new hybrid brain-computer interface that integrates electrical stimulation into imaginary movements based BCI system.Starting from the combination of three aspects of forming a compound paradigm,we are committed to improving the overall performance of the brain-computer interface based on steady-state somatosensory evoked potentials to solve the problem of universality and accuracy of the brain-computer interface,and exploring the effects of steady-state somatosensory evoked potentials on the brain network,The specific work is as follows:Experimental design.The experimental design includes hardware design and experimental paradigm design.In the hardware design,a portable electric stimulator with current stimulation frequency,current magnitude and real-time adjustable stimulation site is designed.The experimental paradigm focuses on the SSSEP paradigm and the SSSEP-MI hybrid paradigm.In the SSSEP paradigm,the Person-specific Resonance-like Frequencies(PRF)of each subject were searched,and the frequency of the Person-specific Resonance-like Frequencies on the EEG signal was studied by using this frequency as the stimulation frequency.In the SSSEP-MI hybrid paradigm,the optimal stimuli in the SSSEP paradigm are selected to study the effect of this paradigm on the classification of features.Feature extraction and feature classification of EEG signals.Firstly,the EEG signal is preprocessed,and then the SSSEP features of the full frequency band and the 1 Hz narrow band in the SSSEP paradigm are extracted according to requirements.The common spatial pattern(CSP)is used for feature extraction,and the support vector machine(SVM)issupported to perform feature classification and then,compares the results with the classification results obtained by the Convolutional Neural Network(CNN).The construction and analysis of brain networks.In This thesis,In this paper,Granger causal analysis method is used to construct the network and study the topological structure of brain network induced by steady-state somatosensory evoked potentials,explore the causal connectivity between various brain regions,and analyze the different brains activated under the state of work tasks and quiet state.The mutual cooperation between the regions promotes the understanding of the working mechanism of the brain and provides theoretical support for the application of the brain-computer interface based on the steady-state somatosensory evoked potential.