Identifying brain dynamic network states via GIG-ICA: Application to schizophrenia, bipolar and schizoaffective disorders

There has been an increasing interest in brain dynamic functional networks revealed by resting-state fMRI data. We hypothesized that dynamic functional networks could offer important information for detecting subtle differences in symptom-related mental diseases. Schizophrenia (SZ), bipolar disorder (BP), and schizoaffective disorder (SAD) have similar symptoms, and there is still controversy about the SAD category. In this paper, we applied a novel method, group information guided ICA (GIG-ICA), to extract functional connectivity states and their fluctuations from dynamic functional network. Using the proposed approach, we analyzed fMRI data of healthy controls, SZ patients, BP patients and two symptom-defined subsets of SAD patients. Results demonstrate that, measured by the dominant functional connectivity state, different groups have a similar pattern, while the two subsets of SAD patients were most correlated to each other, supporting SAD's status as an independent category. The significant difference in the dominant functional connectivity state among these disorders involved cerebellum-related functional connectivity.