The Evaluation Research Of Brain Activity Based On EEG

In recent years, significant advances have been achieved in many fields of the study of brain function, as the intellectual development of infants, the evaluation of cognitive dysfunction, the prevention of encephalopathy like Alzheimer's disease and the monitoring of the brain fag.The method of the study of brain function can be categorized as subjective assessment and objective assessment. By using subjective assessment, rating scales of the study of brain function can be established empirically, such as the rating scales of fatigue, of schizophrenia and of cognitive dysfunction. But subjective assessment can not escape from its own inherent disadvantages of insufficient validity and reliability. With the development of science and technology, more and more objective approaches of studying brain function have appeared, but when it comes to the assessment of brain activity, a quantitative measurable index is still lacking. With regard to that, in this paper, electroencephalogram (EEG) is used in analyzing brain function and, by conducting a system modeling based on a larger number of experiments, an evaluating indicator reflecting brain activity that correspond to specific brain functions is developed. That is expected to provide a quantitative basis for researches of cognitive science and clinical medicine.This paper consists of three phases, as follows:In the first phase, a general introduction of the present research situation, research methods and existing problems of the study of brain function is given. Here, by analyzing advantages and shortcomings of various kinds of research methods, a model of brain activity is brought up based on EEG analysis. Next, a brief introduction, about the mechanism of production, acquisition method and features of the EEG, is made. At last, an overview of time domain, frequency domain, time-frequency analysis and nonlinear dynamic analysis of processing methods of EEG is concluded, and wavelet entropy is adopted to analyze EEG complexity of the depressed patients.The second phase mainly focuses on the study of emotion classification based on EEG (four emotion states:excitement, fear, sleepiness and arousal). The procedure is as follows:(1) Prepare video clips corresponding with those four emotions. (2) Collect subjects' evoked EEG when the video clips are played, and preprocessing of anti-jamming is conducted. (3) Extract the autoregressive (AR) model coefficients, band power, and fractal dimension of the EEG. (4) on principal component analysis (PCA). By using them as characteristic parameters, a lead selection based on mutual information is carried out (finally ten most relevant leads are selected), then conduct a feature reduction based on genetic algorithm (GA). (5) Using Fisher classifiers respectively based on genetic programming and PCA, a research of classification and comparison of those four emotions is conducted, in which the conclusion is drawn that the former recognition rate is higher than that of the latter.In the third phase, a study of brain activity in different affective states is conducted. The procedure of the model of brain activity is as follows:(1) Collect the volunteers'evoked EEG when they experience those four emotions with different stimulus intensities which are categorized as four grades:the nil (in which the brain activity is defined as 0), the weak (in which that is defined as 0.3), the medium (in which that is defined as 0.6) and the strong (in which that is defined as 1). (2) Collect the characteristic parameter of power spectrum and complexity. (3) Build the model of brain activity using the wavelet neural network. Then, according to the defined emotional signals, the paper makes use of wavelet neural network model to train the brain activity and carry out validation and prediction.Taking the fear emotion as an example, we study on modeling of its brain activity. First, video clips of four degrees of fear are prepared, and then the evoked EEG is collected. Last, its brain activity is estimated by the above-mentioned model. The results show the reliability and veracity of the model. EEG analysis methods and the modeling of brain activity, mentioned in this paper, provide a quantitative basis for the further researches of intellectual development, the evaluation of cognitive dysfunction, the prevention of brain disease and the monitoring of the brain fag. Therefore, they have a good application prospect.