Neuroimaging technology has been widely used to explore the biomarkers of mental disorders. There are large individual variations in the brain mechanism of mental disorders, which brings challenges to finding neuroimaging markers of mental disorders. One of the core questions is whether the corresponding and specific brain abnormalities can be found behind the symptom-based diagnostic classification. The speaker will introduce two approaches to this challenge: One approach is to avoid assuming intra-group consistency in patient populations. The speaker will introduce several studies to achieve this approach. Another approach is to detect specific neuroimaging markers by accumulating data. The speaker will introduce the cross-disease neuroimaging feature library established to realize this idea and some findings.