Research On High Performance Mental State Detection Systerm Based On Brain Computer Interface

With the progress of society,people have a higher pursuit of quality of life,and more and more attention is paid to mental diseases such as depression,attention deficit disorder,and movement disorder that seriously affect people’s study,work,and life.However,at present,the diagnosis of mental diseases such as depression mainly relies on the judgment of doctors or self-rating scales.This traditional method has problems of strong subjectivity and low accuracy,which leads to untimely treatment.Aiming at the above problems,this project designs and implements a high-performance mental state detection system based on brain-computer interface,which is used as an auxiliary inspection method to assist doctors in the diagnosis of depression,attention deficit,and movement disorders.The main work and results of this research are as follows:Firstly,this thesis investigates and summarizes the research status of depression,attention deficit disorder and movement disorder at home and abroad,and introduces the neurological basis and analysis methods of scalp EEG signals.Different experimental paradigms are designed to collect the EEG signals of subjects to obtain the experimental data,and decode them according to the communication protocol.In addition,aiming at the problem of many interferences in EEG signals,filtering and independent component analysis are done for the original EEG signals.Then,based on different bands,the features of power spectral density,band energy and average instantaneous energy are extracted as classifier input to apply to different populations and application scenarios.Secondly,in view of the defects of the traditional mental state detection method combined with self-assessment scale,which is not objective enough and easy to be misdiagnosed,this study proposes an objective and high-performance mental state detection method.In this method,one-dimensional convolutional neural network and gated loop unit are connected in series,and attention mechanism is introduced to form a hybrid network model.The one-dimensional convolutional neural network can obtain the local features of the input EEG signal,while the gated loop unit retains the global features of the signal.The attention mechanism can assign different weights to the features extracted by the network,so as to screen out more representative features,reduce the amount of calculation of the network and shorten the training time of the model while ensuring high accuracy.The experimental results show that the mental state detection method proposed in this study has higher classification accuracy,shows good performance on both public data sets and real data sets,and the accuracy of different types of mental state detection has reached more than 96%,which verifies the effectiveness of the method.Finally,starting from the requirements,a mental state detection system based on the Spring Boot application framework is designed and implemented.The system uses the scalp EEG signal collected in real time as the data source,stores it in the database,extracts features,and detects different types of mental states by calling the network model.In addition,the system uses the visualization library Echarts to visualize the EEG signal waveform and detection results,realizing the complete process from EEG signal acquisition and processing to mental state detection and display.The realization of this subject has important research significance and application value for the early diagnosis and adjuvant treatment of some mental diseases.