Schizophrenia MEG Analysis And Classification Based On Hilbert-Huang Transform And Source Localization

Schizophrenia is a psychotic disorder that alters patients' perception, thought processes. It has been found that schizophrenia can also be characterized by a functional abnormality of the brain activity that is reflected in the resting state. With the increasing application of EEG/MEG in studies of brain activity and assisted analysis of neurological disorders, and the more obvious advantages of MEG, using MEG to analyze the brain activity of resting state in schizophrenia will contribute to a better understanding of the nature of schizophrenia.In order to effectively distinguish between normal subjects and schizophrenia, thus to have a clear understanding of the differences between their brain activities. The purpose is achieved based on resting MEG signals from different angles, i.e. the feature extraction, classification and source localization.Firstly, the relevant research methods of MEG signals are introduced, focusing on various algorithms related to feature extraction, classification and source localization, and analyzes and compares the characteristics, advantages and disadvantages.Then, on the basis of the existing methods, a new method based on improved Hilbert-Huang transform and the support vector machine optimized by genetic algorithm is presented. And the classification accuracy of the simulation analysis can achieve 95% by extracting Hilbert weighted frequency as the feature vector, while the classification accuracy through the K-nearest neighbor classifier is 78.33%, whilch illustrates the effectiveness of the method to distinguish between normal subjects and patients with schizophrenia.Finally, a standard low-resolution electromagnetic tomography based source localization method is proposed, which uses the standard low-resolution electromagnetic tomography for initial positioning, and then a comprehensive analysis of the results of the cross-correlation coefficient and partial coherence will be used for further positioning, and the distribution of cortical source time series of normal subjects and patients will also be compared. The final simulation results show that the impact of schizophrenia on the frontal and occipital regions is larger, especially in the frequency range of 4 ~ 13 Hz. This indicates that the source localization method is suitable for the identification of schizophrenia.