Study Of The Functional Network Based On FMRI Data

Schizophrenia is a serious mental disorder, the common symptoms of which include false beliefs, unclear or confused thinking, auditory hallucinations, reduced social engagement and emotional expression,and inactivity. The underlying neuropathological mechanisms of schizophrenia still remain largely unknown, After the hypothesis of neural network ‘disconnection’ which assumes that schizophrenia can arise from the dysfunctional integration of a distributed network of brain regions was proposed, many neuroimaging studies have provided consistent evidence of ‘dysfunctional connectivity’ among the brain regions in schizophrenia patients.Unlike commonly used approaches, we directly constituted the network at the voxel level, this method is applied to this circuit to determine the exact location of functional abnormalities. In our study, the voxels were extracted using resting-state fMRI data from 160 schizophrenia patients and 151 healthy controls. This resulted in connectivity networks with 47,636 nodes and over 1.1 billion links. After that, a two-sample t-test was then performed for all links to reveal the differences across two groups. Only significant voxels(FDR-corrected)which clustered into 11 regions in space were reserved for furtheranalysis. This clusters of voxels with altered functional connectivity in schizophrenia were associated with the regions of default mode network(DMN), including precuneus and cingulate; and attention network mainly including inferior frontal gyrus; thalamus and basal ganglia. Most links that showed a significant change had a decrease in functional connectivity in subjects with schizophrenia especially the connection between regions of precuneus and inferior frontal gyrus. Then,we investigated whether the emerging pattern of dysconnectivity correlated with clinical variables and applied a support vector machine(SVM) approach using the alterations in the clusters as a biomarker to test how well this could discriminate patients from the healthy controls.Finally, we also constituded the functional connectivity at ROI level to compare the results. Parcellation-based analysis averaged out the fine details over small spatial scales, the connection within DMN and the region of basal ganglia can’t find using this approach.