Neuroimaging and modeling

Abstract

SummaryThe links between symptomatic phenomenology of psychiatric disorders and their neurobiological pathophysiology are still understood only in fragments. While functional and structural neuroimaging methods have leveraged our knowledge of regional dysfunction in psychiatric disorders, emerging novel approaches more focused on brain networks or neural patterns promise additional forward progress. Activation likelihood estimation (ALE) performs quantitative large-scale aggregation of neuroimaging findings. Resting-state (RS) correlation captures networks of functional relationships between regions in the idling, non-goal-focused brain. In contrast, meta-analytic connectivity modeling (MACM) captures functional coupling between brain regions in the context of experimental paradigms. These methods may furthermore be exploited to provide data-driven parcellations (CBP, connectivity-based parcellation) of larger brain regions into distinct functional modules. Dynamic causal modeling (DCM), in turn, allows for the automatic selection among a set of connectional network models to delineate effective connectivity dynamics during experimental paradigms. Finally, machine learning (ML) allows for the automatic detection and prediction of diagnosis/treatment-response patterns in massive datasets. Capitalizing on this toolbox of computational modeling methods might considerably further psychiatry and thus benefit patients with mental disorders.