The prevalence of mental disorders remains persistently high worldwide, imposing a substantial burden on public health systems. However， as the etiological mechanisms have yet to be fully elucidated， current clinical diagnosis still relies primarily on subjective symptomatic assessment, lacking objective biological criteria. This symptom-based classification system overlooks the biological heterogeneity of the disease， leading to significant differences in brain mechanisms among patients with similar symptoms， which increases the difficulty of precise diagnosis and treatment. Neuroimaging techniques offer new avenues for exploring neurobiological biomarkers. However， current research is largely confined to the population level， and the transition to personalized clinical applications still faces challenges regarding accuracy and generalizability. This review systematically summarizes the key findings from recent years in multimodal neuroimaging regarding structural and functional abnormalities in the brain associated with mental disorders， with a focus on analyzing the current challenges in neuroimaging research on mental disorders and proposing methods and strategies to address these challenges. Although there is currently a significant gap between neuroimaging findings and clinical translation in psychiatry， the integration of multimodal data， the use of artificial intelligence algorithms， and the conduct of large-scale longitudinal cohort studies can gradually lead to the precise diagnosis and treatment of mental disorders.