Analysis And Classification Of Structural Brain Image Based On Linear Mixed Effect Model

It is one of the important methods to explore the structural characteristics of brain and the structural connection network of brain by diffusion tensor imaging.Structural brain image data has been widely used in the research of personality exploration.The structural brain image data obtained from diffusion tensor imaging are widely used in the exploration and research of personality.In the field of computer,the complex network theory can directly explain the corresponding results of various brain imaging models,which is applicable to the study of human brain.It can spatially divide human brain and study various brain regions.These methods can effectively identify personality traits and accurately diagnose various mental brain diseases.However,in the existing studies,we often neglect the influence of age and gender on brain structure image data in reality,ignoring the possible clustering characteristics of data,which leads to the reduction of the validity of feature selection and ultimately the reduction of classification accuracy.Previous studies have used multivariate linear correlation method to consider the impact of multiple confounding factors on data.However,the method of multivariate linear correlation analysis ignores the non-independence of data,and can not accurately estimate and hypothesis test the influencing factors.In order to solve these problems,a method of correlation analysis based on linear mixed effect model is proposed in this paper.The main innovative work of this paper is as follows:Firstly,we analyze the structural image data,preprocess the structural data,extract the structural features of brain images,construct the structural network,extract the structural network properties,and study the correlation between personality traits and structural image features.Secondly,we face up to the clustering problem of data sets due to age and gender factors,fully consider the impact of age and gender on brain structure images,and construct a linear mixed effect model.Feature selection is performed by the size of the related significance.Thirdly,feature selection is carried out according to the correlation significance size of the correlation analysis,and the feature set obtained based on this is likely to have redundant features,thus affecting the classification results.To solve this problem,Pearson correlation coefficient between two features is calculated in this study,and redundant features are removed according to the size of correlation coefficient,and obtain the feature set in the final classification.In this paper,a method of analysis and classification of brain structure images based on linear mixed effect model is proposed.The characteristics of brain structure are reflected by DTI scalar indexes and structural network properties.Then,correlation analysis is carried out by considering the two confounding factors age and gender,feature selection was carried out based on the results,and then redundant two to two analysis was carried out to remove irrelevant or repeated features.Support vector machine(SVM)is used to classify personality traits.The classification results show that the method of brain structure image analysis and classification based on linear mixed effect model greatly improves the accuracy of personality trait identification.This paper focuses on the study of the characteristic parameters of structural brain image data,the construction of structural connection network and property analysis,and build a linear mixed effect model to analyze the correlation.It is an international frontier basic science problem and also a great need for China to explore personality traits and seek for their imaging symbols,and then create an auxiliary identification model based on this.