Differentiate Negative and Positive Schizophrenia Using Support Vector Machine

Abstract

In the present study, we identified the correlative pattern of gray matter distribution that best discriminates between positive and negative schizophrenia patients, which might provide additional information to psychiatric diagnostic system for mental disorders. First, we applied the voxel-based morphometry (VBM) to compare the gray matter distribution between negative and positive schizophrenia patients. Second, we trained the support vector machine (SVM) to obtain a classification model that classified 20 positive and 11 negative schizophrenic patients. The results showed that 84% subjects were correctly classified. We demonstrated that the united method of VBM and SVM would provide a useful tool for clinical diagnostic systems.