The Application Of Genetic-Evolutionary Random SVM Cluster On The Magnetic Resonance

The brain is still a delicate and mysterious organ for human beings,and the exploration of the brain has attracted the attention of governments around the world and the entire academic community.In recent years,due to the rapid development of brain imaging technology,finding abnormal functional connections and lesions of brain diseases is an important research direction of current brain science.This study is based on the support vector machine(SVM)method to construct genetic-evolutionary random SVM cluster,and use the model to study two common brain diseases,Autism Spectrum Disorder(ASD)and Asperger’s Syndrome(AS).The main contents of this paper are as follows:(1)The genetic-evolutionary random SVM cluster is proposed.SVM classifier is used as the basic classifier of the cluster.The samples and features are randomly selected to construct the cluster.Then the cluster is evolved by using the idea of genetic algorithm,and the optimal feature selection of the samples is gradually achieved.Because the new method uses cluster to overcome the performance fluctuation of single classifier,and uses genetic evolution method to optimize the cluster interior,these measures effectively improve the generalization performance of the cluster.(2)The genetic-evolutionary random SVM cluster was used to study Autism Spectrum Disorder.The fMRI data of 103 ASD patients and 106 Healthy Controls(HCs)were collected from the ABIDE database.The brain functional connectivity network was established and the graph theory metrics of brain regions were used as sample features.The brain functional connectivities of these subjects were analyzed by using the proposed model.The result found that the genetic-evolutionary random SVM cluster can effectively identify Autism Spectrum Disorder patients from HC and achieve the highest classification accuracy of 96.8%.The abnormal functional connections of ASD patientsin superior limbic gyrus,inferior frontal gyrus,fusiform gyrus and hippocampus could be found.The experimental results show that this method may be an assistant means of autism diagnosis and provide a new technical route for the study of autism.(3)The genetic-evolutionary random SVM cluster was used to study Asperger’s Syndrome.The resting state fMRI data of 62 AS patients and 86 HCs were collected from the ABIDE database.By constructing functional connectivities as the input features and using the proposed method to classify AS patients from HCs.As a result,the average accuracy of the model reached 83.95%.At the same time,the angular gyrus,anterior cuneiform lobe,caudate nucleus and cuneiform lobe can be found as important regions which effectively classified AS patients and HC.This study proves that the model has good generalization performance and can provide important reference for the diagnosis and treatment of Asperger’s disease.