HELIOS: A spatial omics-informed multi-organ cell atlas enables AI-based tumour microenvironment profiling and biomarker discovery from routine histopathology

HELIOS transfers spatial-omics cell identities into routine H&E, integrating 12.9 million cells from 60 specimens across 13 tissue types to enable AI profiling of tumour microenvironments.

• Why it matters: Spatial omics is high-resolution but hard to scale across large clinical cohorts, limiting interpretable, clinically usable microenvironment biomarkers from routine pathology.
• What they did: HELIOS uses spatial-omics-derived supervision to train a multiclass detector on H&E, incorporating 12.9 million cells from 60 specimens across four spatial profiling platforms and 13 tissue types.
• The result: In colorectal, breast, and lung cancers, HELIOS recapitulates molecular-subtype microenvironment states and links tumour-infiltrating lymphocytes with favourable outcomes, while neutrophil and fibroblast phenotypes show cancer-specific prognostic associations.

AI summary of the abstract · may contain errors · not medical advice · read the paper.