Discriminating schizophrenia and schizo-obsessive disorder: a structural MRI study combining VBM and machine learning methods

Abstract

Schizo-obsessive disorder is characterized by the clinical syndrome in which comorbid obsessive–compulsive disorder accompanies schizophrenia. A substantial number of studies have investigated the neuropsychological and clinical differences between schizophrenia and schizo-obsessive disorder. However, the neurostructural differences between these two groups have not been adequately investigated. The aim of this study was to explore gray matter differences between schizophrenia and schizoobsessive patients using voxel-based morphometry and support vector machines combined with feature selection algorithm. Twenty-three schizophrenia and 23 schizo-obsessive patients matched by age, gender and handedness were recruited. Clinical assessments were completed in addition to high-resolution structural MRI scanning. Group differences were investigated using contrast maps, and significant regions were subjected to a feature selection and support vector machine hybrid model. In addition, voxelof- interest values for the commonly shared brain areas between schizophrenia and OCD reported in previous meta-analyses were also used as inputs in this step. The results showed that schizo-obsessive patients had greater gray matter densities in paracentral areas (including supplementary motor area) and middle cingulate gyrus than schizophrenia patients. These brain areas together with the fronto-subcortical areas could successfully discriminate two groups with an accuracy of 78.26 %. Our results provide the first neuroanatomical evidence that schizo-obsessive disorder and schizophrenia may be two distinct clinical entities. Based on these findings, considering schizo-obsessive disorder as a subtype of schizophrenia is discernible.