Understanding the heterogeneity of mental disorders remains a major challenge in neuroscience. Individualized modeling approaches have emerged to overcome the constraints of traditional group-level paradigms by characterizing subject-specific neural alterations. However, current individualized brain network constructions predominantly focus on capturing subject-specific abnormalities, which neglects common group-level characteristics underlying phenotypic similarities within the same diagnosis. In this study, we propose a novel structural-functional individual brain network (SFINe) approach that integrates structural heterogeneity—capturing subject-specific deviations in gray matter volume—with functional commonality—modeling common stable connectivity patterns across individuals. Specifically, normative modeling and multi-group perturbation based individualized differential structural brain network (MPIDS) are used to construct individualized structural networks, while seed-based functional connectivity is used to identify group-level functional patterns. Functional networks are guided by regions exhibiting individual-specific structural deviations, thereby aligning functional commonality with anatomical heterogeneity. Results show that the proposed SFINe consistently achieves the best performance in both symptom prediction and diagnostic classification tasks compared with four alternative methods across two independent autism datasets and three brain atlases. These highlight the potential of integrating structural heterogeneous features and functional common information to comprehensively characterize individual-specific brain network alterations, thereby providing a new perspective for investigating the neural substrates underlying mental disorders.