Neuroimaging in schizophrenia: From group-average abnormalities to individualised circuit models

This narrative review synthesises advances in multimodal neuroimaging that, over the past five years, have substantially refined models of schizophrenia pathophysiology. Converging evidence from structural, diffusion, functional, molecular, and computational studies challenges static, regionally focal, or single-neurotransmitter accounts of the disorder. Instead, contemporary findings support a framework in which schizophrenia reflects heterogeneous, developmentally anchored deviations in distributed neural circuits that manifest as large-scale network instability. Structural and diffusion imaging reveal individually variable patterns of cortical, white matter, and thalamic subnuclei alterations. Functional MRI demonstrates impaired regulation across salience, default mode, executive, and thalamocortical systems, while positron emission tomography and magnetic resonance spectroscopy implicate subregion-specific dopaminergic and glutamatergic abnormalities. Computational modelling further indicates that these multilevel disturbances may converge on altered synaptic gain and excitation–inhibition imbalance within cortical microcircuits, providing a mechanistic substrate for systems-level dysconnectivity. Importantly, heterogeneity is not incidental but central: normative modelling and biologically informed subgrouping demonstrate that while group-level effects are robust, the anatomical and neurochemical loci of deviation vary substantially across individuals. We propose that shared network instability may arise from diverse developmental perturbations across cortico–striato–thalamo–cortical circuits. Future progress will depend on longitudinal, harmonised, and multimodal study designs capable of modelling individual trajectories across risk stages and treatment exposure. Conceptualising schizophrenia as a dynamically evolving circuit disorder therefore offers an integrative framework that bridges molecular dysfunction and clinical expression and provides a roadmap for mechanism-informed stratification, while clinical translation remains a longer-term objective.