The Research Of Detection And Evaluation Of Motor Function Disorder Using EEG

Motor function disorders is an important manifestation of various brain injury diseases,such as epilepsy,stroke and other diseases.The detection and evaluation of motor function in patients is an important method in the process of diagnosis and rehabilitation.At the same time,the EEG signal is a biological signal that directly reflects the activity of the brain.Through the analysis of EEG signals,we can obtain a lot of physiological and pathological information.The aim of this paper is to evaluate the motor function of patients with brain injury by analyzing EEG.However,it is a complex project for analyzing EEG signals efficiently.In this paper,we choose epilepsy patients with motor function impairment as the subject in this study.The main purpose of this study is extracting and identifying the epileptic features of the patients in the EEG signals.In the process of the study,this paper focuses on the algorithms of EEG signal processing,including the denoising method,feature extraction method and pattern recognition algorithm of EEG signals.The main contents of this paper include the following aspects:(1)This paper proposes a denoising method of EEG based on Denoising Source Separation(DSS).DDS is a new method in Blind Source Separation,which can design exclusively a suitable denoising function according to the characteristics of the EEG.After the signals are processed by DSS,we can obtain the source signals.In this paper,firstly we design a simulation experiment to select the denoising function of DDS based on the epilepsy EEG.Then,source separation processing and denoise processing are performed by the DSS.The results of simulation and real EEG signal processing show that the denoising effectiveness of proposed method is superior to that of the method using the Independent Component Analysis.(2)A novel feature extraction method of EEG using multiple entropies fusion is proposed.Since a single entropy only can express the complexity of signal from one view and often lack the overall measurement.In order to achieve more comprehensive expression from multiple perspectives,an EEG feature extraction method based on entropies fusion of different physical meanings is proposed to improve the expression of EEG features.This innovative approach has been published in Neural Computing & Applications(SCI journal).This paper firstly analyzes the physical meaning and expression of four different entropies in EEG.Then we employ the experiment to analyze the improvement of EEG signal classification on different entropies fusion.(3)This paper proposes an ensemble classifier based on Extreme Learning Machine(ELM).ELM is a new classifier and one of the hotspots of EEG classifier design in recent years.However,conventional ELM has the poor generalization ability and unstable classification results.To solve these problems,in this paper,we design an ensemble classifier that the multiple ELM classifiers are integrated into a strong classifier using Bagging and Adaboost.The experiments’ results show that the classifier designed in this paper can not only obtain better classification results,but also solve the problem of classification instability in ELM classifier.