Research2 sources
Bridging morphology and molecular data for foundation models in oncology
Nature · By Eric Deutsch ·
The Brief · written from 2 sources
- A study in Nature Cancer introduces THREADS, a multimodal foundation model trained to align histology with molecular data in oncology.
- The authors (Vaidya et al.) report that THREADS is driven by tissue molecular composition features.
- The sources report high efficiency and generalizability of the model.
Why it matters
It suggests that tissue images and molecular data can be linked in one model, which may help computational pathology research connect morphology with genomics.
Who’s affected
- oncology researchers
- computational pathologists
- #oncology
- #computational pathology
- #foundation models
- #multimodal learning
- #genomics
Nature Cancer, Published online: 20 November 2026; doi:10.1038/s43018-025-01092-0 Foundation models are reshaping computational pathology. A study introduces THREADS, a multimodal system trained to align histology with molecular data, demonstrating that morphology and genomics…