Revamping neuroimaging analysis to reveal biomarkers of adolescent mental health

Abstract

Advances in neuroscience research provide an unprecedented opportunity to identify the etiopathogenesis of mental health disorders. Yet it has proven difficult to find reliable associations between neurobiological phenotypes and real-world mental health experiences, particularly among youth. This Perspective addresses two pervasive assumptions inherent to many functional neuroimaging studies that diminish the predictivity of the data. First, studies assume that aligning data across individuals on the basis of the anatomy of the brain is sufficient to align their brain function. Individual brains vary meaningfully in the localization of functions, particularly across development and in clinical populations; neglecting this variability in functional neuroanatomy risks washing out rich and reliable patterns of individual-specific information. Second, studies assume that the underlying signal embedded in brain measurements over space and time can be modeled with simple transformations from high dimensions (that is, voxels) to low or single dimensions (that is, regional averages). However, the latent structure of brain activity and behavior is often complex and nonlinear. To overcome these assumptions, we suggest alternative methodological approaches that have yielded novel insights into the neurobiology of cognition and mental health symptoms in adolescence. Building robust predictive models of psychiatric problems requires methodology that can capture the richness and complexity of the brain and behavior. Neuroscience research struggles to link neurobiological phenotypes with real-world mental health experiences, especially in youth. Here the authors challenge assumptions in functional neuroimaging studies, proposing alternative methods that reveal complex, individual-specific brain patterns, enhancing predictive models of psychiatric issues and advancing the understanding of adolescent mental health.