A Default Mode Network Whiter Matter Connection Study Of Schizophrenia,Bipolar Disorder And Major Depression

Background: Schizophrenia,bipolar disorder and major depression are the most common three mental disorders in psychiatry.The symptoms and diagnoses of these mental disorders often overlap in clinical practice.In recent years,diagnostic methods including neuroimaging have been used in accessory diagnosis of mental disorders.Functional and structural connections abnormalities of different mental disorders have been found through neuroimaging.There are both common connection changes and unique characteristics among different mental disorders,indicating the possibility of making diagnosis through neuroimaging connection characteristics.The relative stability of structural connections in the white matter of the brain could be help in the differential diagnosis of mental disorders.In this study,probabilistic tractography was used to explore the similarities and differences of default mode network(DMN)white matter connection in schizophrenia,bipolar disorder and major depression and the possibility of accessory diagnosis.Methods: In this study,97 schizophrenia patients,92 bipolar disorder-type I patients,47 major depression patients and 58 healthy controls were recruited.All patients met with DSM-IV diagnostic criteria.DTI images were acquired by MRI scanning,probabilistic tractography was used to evaluate the white matter connections of DMN network.Connection strengths and small-worldness were compared by ANOVA between patients with schizophrenia and bipolar disorder,unipolar depression and bipolar depression.patients’ groups were matched in age and gender.symptom correlations of DMN white matter connections and network properties were compared among the three mental disorders.Diagnostic classification was performed through machine learning.Results: 1.In schizophrenia and bipolar disorder-type I,the connection strengths decreased between the left medial superior frontal gyrus and the right anterior cingulate and paracingulate gyrus(SCH<HC,p=0.0002;BD<HC,p=0.021),and between the bilateral anterior cingulate and paracingulate gyrus(SCH<HC,p=0.0001;BD<HC,p=0.044)as compared with healthy control.Compared with bipolar disorder-type I and healthy control,the connection strengths decreased between the left precuneus and the right posterior cingulate gyrus(SCH<BD,p=0.002;SCH<HC,p=0.0005),the left precuneus and the right angular gyrus(SCH<BD,p=0.025;SCH<HC,p=0.0005),and the bilateral precuneus(SCH<BD,p<0.0001;SCH<HC,p<0.0001)in schizophrenia.There’s no significant difference in network properties between bipolar disorder-type I and healthy controls.In schizophrenia,network properties increased in global efficiency(SCH>HC,p=0.029)and decreased in local efficiency(SCH<HC,p=0.01).2.Compared with the healthy control,unipolar depression and bipolar depression showed an increase white matter connection between the right posterior cingulate cortex and the right supramarginal gyrus(BD>HC,p=0.0005;MD>HC,p=0.0018).There’s no significant difference in the network properties in unipolar depression and bipolar depression compared with the healthy control.The global efficiency of bipolar depression was lower than unipolar depression(BD-I>MD,p=0.012).3.Through machine learning,the average accuracy of classification was 66.7% in schizophrenia and bipolar disorder-type I,67.9% in unipolar depression and bipolar depression.4.Although the global efficiency of DMN in schizophrenia increased,when the DMN communication pathway extended to the whole brain,the global efficiency of schizophrenia was decreased(SCH<BD,P=0.004;SCH<MD,P=0.011;SCH<HC,P=0.022),but no significant difference was found between bipolar disorder-type I,major depression and healthy controls.Conclusions: 1.There may be specific characteristics of DMN brain white matter network in schizophrenia and bipolar disorder-type I,unipolar depression and bipolar depression.2.There may be not only differences in DMN network internal and external nodes and global efficiency among the three different mental disorders,but also common abnormalities such as DMN network internal brain white matter connectivity,local efficiency changes.3.A moderate classification accuracy can be achieved through machine learning in discrimination of schizophrenia and bipolar disorder-type I,unipolar depression and bipolar depression.4.The characteristics of DMN white matter network can be hopefully used as brain-structural signature for the cross diagnosis of schizophrenia,bipolar disorder-type I and depression,which provides a new clue for the accessory diagnosis of mental disorders.