Investigation Of Brain Abnormalities And Neurobiological Subtypes In Schizophrenia Using MRI And Machine Learning Techniques

Part 1: Alterations in whole-brain resting-state effective connectivity network in schizophrenia ?Background and Aims?Numerous neuroimaging studies have revealed that schizophrenia was characterized by wide-spread dysconnection among brain regions during rest measured by functional connectivity(FC).In contrast with FC,effective connectivity(EC)provides information about directionality of brain connections and is thus valuable in mechanistic investigation of schizophrenic brain.However,a systematic characterization of whole-brain resting-state EC and how it captures different information compared with resting-state FC in schizophrenia are still lacking.The aim of this study was to systematically characterize the abnormalities of EC,compared with FC,in schizophrenia,and to test its discriminative power as a neuroimaging marker for schizophrenia diagnosis.?Methods?Whole-brain EC and FC networks were constructed using resting-state f MRI(functional magnetic resonance imaging)data and compared between 103 patients with schizophrenia and 110 healthy participants.Pattern classifications between patients and controls based on whole-brain EC,whole-brain FC or their combination were further performed using multivariate pattern analysis.?Results?We identified 17 EC significantly disrupted(mostly decreased)in patients,among which all were associated with the thalamus and 15 were from limbic areas(including hippocampus,parahippocampus and cingulate cortex)to the thalamus.In contrast,abnormal FC were widely distributed in the whole brain.The classification accuracies for distinguishing patients and controls using whole-brain EC and FC patterns were 78.6% and 82.7%,respectively,and was further improved to 84.5% when combining EC and FC.?Conclusions?Schizophrenia is featured by disrupted ‘limbic areas-to-thalamus' EC,in contrast with diffusively altered FC.Moreover,both EC and FC contain valuable and complementary information which may be used as diagnostic markers for schizophrenia.Part 2: Investigation of abnormalities of the thalamus and its diagnostic value in schizophrenia using multimodal neuroimaging and machine learning techniques ?Background and aims?The thalamus,a critical hub in bidirectional information transfer between different cortical areas as well as between cortical and subcortical areas,plays an important role in cognitive functions and emotional processing.Findings from previous studies and also from Part I of this thesis indicate that structural and functional abnormalities in thalamus may be a characteristic of schizophrenia.However,relatively few neuroimaging studies on schizophrenia have focused specifically on the thalamus,and consequently a systematic characterization of the structural and functional abnormalities of the thalamus in schizophrenia is still lacking.Hence,we aimed to explore the structural and functional changes of the thalamus using multimodal neuroimaging in patients with schizophrenia,and to assess its diagnostic value using multivariate pattern analysis(MVPA).?Methods?Multimodal neuroimaging data were acquired from 97 patients with chronic schizophrenia and 105 healthy controls.Several neuroimaging indices were calculated within bilateral thalamus and compared between patients and healthy controls at voxel level,including functional metrics such as amplitude of low-frequency fluctuations(ALFF),regional homogeneity(Re Ho)and cerebral blood flow(CBF)and structural metrics such as gray matter volume(GMV)and white matter volume(WMV).Then,the relationship between these neuroimaging indices and clinical measures were examined using Pearson‘s correlation coefficient.Last,MVPA based on these multimodal neuroimaging indices were performed to classify between patients with schizophrenia and healthy controls.?Results?Significantly decreased ALFF,Re Ho,GMV and local WMV,as well as increased CBF and local WMV were detected in thalamus in patients with schizophrenia.In addition,WMV was found to be decreased in some subregions and increased in some other subregions within the thalamus.Correlation analyses revealed that only GMV showed significant correlations with multiple clinical measures,such as illness duration,scores of several subscales(positive symptom,activation,bigotry)of the Positive and Negative Syndrome Scale(PANSS),as well as auditory hallucination duration.Classification analyses showed that all these neuroimaging indices could successfully distinguish patients from healthy controls(accuracy: 60.40%-70.31%,all significantly higher than chance level).Moreover,the fusion of these neuroimaging indices considerably improved the classification accuracy(78.69%).?Conclusions?Multimodal neuroimaging confirmed the structural and functional abnormalities of the thalamus in schizophrenia,and showed a significant correlation between GMV and several clinical measures.These results provide evidence for the pathophysiology of schizophrenia.Successful classification between patients and controls based on these thalamic changes,especially their combination,suggests their potential diagnostic value for schizophrenia.Part 3: Subtyping schizophrenia based on the spatial patterns of gray matter volume in the brain ?Background and Aims?Although it has become a consensus that abnormal brain structure and function exist in patients with schizophrenia,different brain imaging studies often report different brain abnormalities.One important reason for such inconsistency between studies may be the heterogeneity of this mental disorder.However,subtypes of schizophrenia have not been clearly identified.The aim of the present study is therefore to explore the neurobiological subtypes of schizophrenia based on GMV.?Methods?MRI data were acquired from 97 patients with chronic schizophrenia and 105 healthy controls and voxel-wise GMV were calculated for each participant.K-means clustering analysis was performed to identify subtypes of the patients with schizophrenia based on their spatial patterns of GMV.To identify the characteristics of each subgroup,comparisons of GMV and FC were performed between different subgroups of patients as well as between each patient subgroup and healthy controls.Clinical measures were also compared between different patient subgroups.?Results?We identified two patient subgroups based on GMV pattern,each showing different characteristics of alterations in brain structure(i.e.,GMV)and function(i.e.,FC)and of clinical symptoms.In terms of brain structure,Subgroup 1 showed local increase of GMV mainly in the left globus pallidus and cerebellum,whereas Subgroup 2 showed extensive decrease of GMV mainly in the prefrontal,temporal and occipital areas,insula,anterior and middle cingulate,thalamus and caudate nucleus.In terms of brain function,the two patient subgroups did not show significantly different FC;nevertheless,when compared with healthy controls,Subgroup 2 showed more extensive FC alterations relative to Subgroup 1.In terms of clinical measures,Subgroup 2 showed more severe negative symptoms compared with Subgroup 1.?Conclusions? We characterized the heterogeneity of patients with chronic schizophrenia based on their GMV,and revealed two subtypes showing different profiles in brain structure,function and clinical symptoms.These results provide evidence for the heterogeneity of the neural mechanisms underlying chronic schizophrenia.