Scalable topic modelling decodes spatial tissue architecture for large-scale multiplexed imaging analysis

Recent progress in multiplexed tissue imaging is deepening our understanding of tumor microenvironments related to treatment response and disease progression. However, analyzing whole-slide images with millions of cells remains computationally challenging, and few methods provide a principled approach for integrative analysis across images. Here, we introduce SpatialTopic, a spatial topic model designed to decode high-level spatial tissue architecture from multiplexed images. By integrating both cell type and spatial information, SpatialTopic identifies recurrent spatial patterns, or "topics," that reflect biologically meaningful tissue structures. We benchmarked SpatialTopic across diverse single-cell spatial transcriptomic and proteomic imaging platforms spanning multiple tissue types. We show that SpatialTopic is highly scalable to large-scale images, along with high precision and interpretability. It consistently identifies biologically and clinically significant spatial topics, such as tertiary lymphoid structures, and tracks spatial changes over disease progression. Its computational efficiency and broad applicability will enhance the analysis of large-scale imaging datasets.