Human Brain Structural Networks Of Patients With Comorbid Depression And Anxiety Disorders

Although depression and anxiety disorders are two independent diseases,they often coexist with each other,which is called "comorbidity" in medicine,depressive patients with anxiety(referred as "comorbid patients")have many features,such as late age of onset,severe illness,high suicide risk: Based on diffusion tensor imaging(DTI)of the human brain,we had studied topological properties of structural brain networks of depressive patients with anxiety patients by using the methodologies of complex network theory and machine learning,quantitative analysis the relationship topological properties of structural brain networks and he severity of the disease(degree of depression or anxiety)of depressive patients with anxiety,to explore of the neuropath logical mechanisms of disease had produced effect on structural brain networks.The whole cerebral cortex was divided into 1024 regions by the anatomical label map(ALL),Fiber tracking was performed in the whole cerebral cortex of each subject to reconstruct white matter tracts of the brain using the fiber assignment by continuous tracking algorithm(FACT),which deterministic tractography were applied to map the white matter structural networks.In that way,we had analyzed depressive patients with anxiety and depression patient as well as normal control topological properties of structural brain networks.The result mainly showed that: The structural brain networks in three groups showed small-world properties and highly connected global hubs predominately from association cortices;Significant differences of network properties(clustering coefficient,characteristic path lengths,local efficiency,global efficiency)were found between depressive patients with anxiety and depression groups;depression group showed significant changes of nodal efficiency in the brain areas primarily in the temporal lobe and bilateral frontal gyrus,compared to depressive patients with anxiety and normal control groups.We had studied on the correlation between the local topological properties of the structural brain networks in depression and the severity of the disease.The result mainly showed that: there was no significant correlation between the global properties of the depression comorbid with anxiety group and the severity of disease;The brain structural networks were constructed using the fiber number(FN),significant negative correlation was found between the global efficiency?characteristic path lengths?mean of anatomy distance and the total scores of HAMA-14 and significant negative correlation was found between the modularity and total scores of SDS in the topological properties of depression group(FN threshold of 3,4,5),in the depression group.The brain structural networks were constructed using the Fractional Anisotropy(FA),significant negative correlation was found between the local efficiency,characteristic path lengths,mean of anatomy distance and the total scores of HAMA-14 in the topological properties of depression group(FA threshold of 02,0.25,0.3).Significant negative correlation was found between the modularity and total scores of HAMD-17 in the topological properties of depression group(FA threshold of 0.25).We had studied on the significant correlation between the node efficiency properties of the structural brain networks in depression and the severity of the disease.Significant correlation was found b between the severity of the disease and the nodal efficiency properties of in the brain areas primarily in the(superior frontal gyrus,insula and caudate nucleus etc)in the depression comorbid with anxiety and depression groups.We had studied the application of machine learning theory on feature extraction method in topological properties of structural brain networks data.we had found that the methodology of SVM-RFE feature selection showed best performance.For example,in the classification test of depressive patients from depressive patients with anxiety,we had used the data combing nodal efficiency of feature and applied the method of SVM-RFE feature selection;we had achieved the excellent results of that the specificity=95%,sensitivity=94.4% and accuracy =94.7%.All in all,in this thesis,we had analyzed changes of topological properties of structural brain networks of depressive patients with anxiety,differences were analyzed by multiple perspectives.With the objective of clinical psychiatry scale verified the clinical availability of methods used.The findings in the thesis may provide reliable reference for the neuropath logical mechanisms of depressive patients with anxiety,and provide a new potential biomarker for clinical diagnosis and early intervention of depressive patients with anxiety.