Abstract
OBJECTIVES: Schizophrenia (SCZ) is associated with disrupted functional brain connectivity, and antipsychotic medications are the primary and most commonly used treatment for schizophrenia. However, not all patients respond to antipsychotic medications. METHODS: The study is aimed at investigating whether the graph-theory-based degree centrality (DC), derived from resting-state functional MRI (rs-fMRI), can predict the treatment outcomes. rs-fMRI data from 38 SCZ patients were collected and compared with findings from 38 age- and gender-matched healthy controls (HCs). The patients were treated with antipsychotic medications for 16 weeks before undergoing a second rs-fMRI scan. DC data were processed using DPABI and SPM12 software. RESULTS: SCZ patients at baseline showed increased DC in the frontal and temporal gyrus, anterior cingulate cortex, and precuneus and reduced DC in bilateral subcortical gray matter structures. However, those abnormalities showed a clear renormalization after antipsychotic medication treatments. Support vector machine analysis using leave-one-out cross-validation achieved a correct classification rate of 84.2% (sensitivity 78.9%, specificity 89.5%, and area under the receiver operating characteristic curve (AUC) 0.925) for differentiating effective subjects from ineffective subjects. Brain areas that contributed most to the classification model were mainly located within the bilateral putamen, left inferior frontal gyrus, left middle occipital cortex, bilateral middle frontal gyrus, left cerebellum, left medial frontal gyrus, left inferior temporal gyrus, and left angular. Furthermore, the DC change within the bilateral putamen is negatively correlated with the symptom improvements after treatment. CONCLUSIONS: Our study confirmed that graph-theory-based measures, combined with machine-learning algorithms, can provide crucial insights into pathophysiological mechanisms and the effectiveness of antipsychotic medications.