A Multimodal Foundation Model of Spatial Transcriptomics and Histology for Biological Discovery and Clinical Prediction
This work addresses the problem of high costs and low throughput in spatial transcriptomics for biomedical researchers and clinicians by providing a scalable framework for spatially informed discovery and clinical precision medicine, representing a novel method for a known bottleneck.
The researchers tackled the challenge of integrating costly spatial transcriptomics with widely available histology images by developing STORM, a foundation model trained on 1.2 million profiles across 18 organs, which outperforms existing methods in predicting spatial gene expression from H&E images across 11 tumor types and improves immunotherapy response prediction in 7,245 patients.
Spatial transcriptomics (ST) enables gene expression mapping within anatomical context but remains costly and low-throughput. Hematoxylin and eosin (H\&E) staining offers rich morphology yet lacks molecular resolution. We present \textbf{\ours} (\textbf{S}patial \textbf{T}ranscriptomics and hist\textbf{O}logy \textbf{R}epresentation \textbf{M}odel), a foundation model trained on 1.2 million spatially resolved transcriptomic profiles with matched histology across 18 organs. Using a hierarchical architecture integrating morphological features, gene expression, and spatial context, STORM bridges imaging and omics through robust molecular--morphological representations. STORM enhances spatial domain discovery, producing biologically coherent tissue maps, and outperforms existing methods in predicting spatial gene expression from H\&E images across 11 tumor types. The model is platform-agnostic, performing consistently across Visium, Xenium, Visium HD, and CosMx. Applied to 23 independent cohorts comprising 7,245 patients, STORM significantly improves immunotherapy response prediction and prognostication over established biomarkers, providing a scalable framework for spatially informed discovery and clinical precision medicine.