Mice, as a core model organism in neuroscience, are undergoing a technological transition in whole-brain atlas construction, shifting from traditional anatomical approaches to multidimensional molecular-level analysis. This marks a new phase in brain research methodology characterized by higher resolution and systemic integration. Spatial transcriptomics technologies have significantly advanced the biological depth of neuroscience studies, offering novel paradigms for exploring dynamic neural circuit evolution and cellular diversity in the brain. By integrating traditional anatomical localization, single-cell molecular connectomics, and functional imaging for macroscopic dynamic tracking, brain atlas research achieves "molecule-circuit-behavior" tri-level integration, constructing molecular regulatory networks underlying dynamic neural circuit remodeling. However, current challenges persist in technical integration. Reference brain atlases hold great promise for elucidating brain homeostasis mechanisms, identifying abnormal circuit metabolic features in neurological disorders (e.g., anxiety), and screening therapeutic targets. Future brain atlas research must advance multimodal technology fusion and cross-dimensional data integration to achieve precise mapping from static structures to dynamic functional networks, providing revolutionary tools for neuroscience.