Human neuroimaging has been a core component of both research in psychiatry and conceptual models of the brain circuit-level mechanisms underlying psychopathology. Despite landmark neuroimaging research over the past 25 years, we still lack the level of precision and insight needed for bringing neuroimaging tools into clinical care contexts. This brief review examines historical research trends in psychiatric neuroimaging, as well as the basic assumptions underlying current efforts, in order to understand factors that have limited the impact of neuroimaging efforts thus far. These factors include the pitfalls of case-control designs, confounders inherent in associational research approaches, and the challenges in embracing fully data-driven analyses. Several critical gaps emerge, the addressing of which could provide the critical new insights that have long been sought from neuroimaging. These include transitioning from group-to individual-level analyses (and through this to intervention studies carried out robustly at the level of individual patients), building "big data" from a longitudinal perspective and not only a cross-sectional one, a greater focus on identifying causal mechanisms, and the development of tools such as electroencephalography in addition to the dominant MRI methods to aid translation to real-world clinical care. Despite the still-unrealized potential of psychiatric neuroimaging, there is now much to be excited about as previous learnings are converted into fundamentally new directions. Indeed, we may now be at an inflection point for neuroimaging if the typical study designs are left in the past and the field systematically and thoughtfully embraces the challenges that the past 25 years of research have now made apparent.
Keywords: Neuroimaging; Psychopathology.