Research And Application Of Human Brain Based On Complex Network

Autism is a more and more common mental illness in the children's group.Autism seriously afflicts the lives of children and hinders the healthy growth of children,and it imposes a heavy burden on their families and society.If left untreated,the child will lose the ability to communicate with others.Therefore,it is especially important to do early prevention and early treatment of autism.More and more studies have shown that there are certain problems in the brain development of patients with mental illness.If we can quantify some differences between the brain of the patient and the normal human brain,then the diagnosis and treatment of the disease can be better.Based on the nuclear magnetic resonance data of autism patients,this paper analyzes the brain network by using relevant theoretical knowledge of complex networks.The work done in this article can be described as follows:(1)In this paper,the structural and functional networks of autistic patients and normal controls are constructed,and the differences between the two networks are analyzed by using complex networks and graph theory.The results show that the small-world attribute of the structural and functional networks of autistic patients were degraded to a certain extent.When the network is faced with attack,the structural network of normal people shows a slightly better recovery ability.In addition,we find that the betweenness of structural network of autistic patients increased abnormally in thalamus and other regions,which also explains the abnormal sensory sensitivity of autistic patients.According to the analysis of network synchronization,we find that the functional network of autistic patients is closer to the regular network than that of normal people,so there is a problem of growth retardation.In addition,the level and modularity of normal people's functional network are higher,so they have better information processing ability.(2)The paper analyzes the resting brain activity signals,and mainly analyzes the fALFF signal,ReHo signal and VMHC value.The results show that the fALFF and ReHo signals in the brain of autistic patients increase abnormally in the paracentral lobule and the supplementary motor area,which indicates that the motor function of autistic patients increases abnormally.In the middle frontal gyrus,both fALFF and ReHo signals are weakened in autistic patients,which indicates that the short-term memory ability of autistic patients has deteriorated.In addition,the VMHC value of paracentral lobule and complementary motor area in autistic patients is abnormal,which is consistent with the results of the previous two signal analysis.The symmetry of brain function in normal subjects of different age groups has also been analyzed.The results show that the symmetry of brain function would decline and the division of labor between left and right hemispheres would become clearer after the age of 27.(3)In this paper,the Granger causal effect is used to construct the brain effect network,and the clustering coefficient of the effect network of the two groups is analyzed.The results show that there is no significant difference between the two groups.In addition,by analyzing the consistency between the structural networks of the two groups and the existence of the respective functional networks,it is found that the brain connection of the patient group does not have obvious abnormality.In the analysis of the consistency between the two networks of three groups of normal people of different ages,it is found that the consistency does not change significantly with the increase of age,which indicates that there is a certain synchronization between the structural network and the functional network from development to aging.In addition,the differences of clustering coefficients and local efficiency between the three groups of subjects are analyzed.The results show that the topological changes of the functional network do not synchronize completely with the topological changes of the structural network.There is a certain degree of independence between these two.In addition,it is found that the performance of the normal brain network reaches a relatively high level between the age of 18 and 23.(4)In this paper,the abnormal results found in the above analysis are taken as features,then we use multiple machine learning classification algorithms such as SVM and LR,and several data dimensionality reduction and feature selection methods to improve generalization performance.Finally,after model integration,the classification accuracy rate reaches 86.24%.The precision and specificity have also reached more than 80%,which can be basically applied to the auxiliary diagnosis of autism.