Abstract
Greater understanding of individual biological differences is essential for developing more targeted treatment approaches to complex brain disorders. Traditional analysis methods in molecular imaging studies have primarily focused on quantifying tracer binding in specific brain regions, often neglecting inter-regional functional relationships. In this study, we propose a statistical framework that combines molecular imaging data with perturbation covariance analysis to construct single-subject networks and investigate individual patterns of molecular alterations. This framework was tested on [18F]-DOPA PET imaging as a marker of the brain dopamine system in patients with Parkinson's Disease (PD) and schizophrenia to evaluate its ability to classify patients and characterize their disease severity. Our results show that single-subject networks effectively capture molecular alterations, differentiate individuals with heterogeneous conditions, and account for within-group variability. Moreover, the approach successfully distinguishes between preclinical and clinical stages of psychosis and identifies the corresponding molecular connectivity changes in response to antipsychotic medications. Mapping molecular imaging networks presents a new and powerful method for characterizing individualized disease trajectories as well as for evaluating treatment effectiveness in future research.