GigaTIME

GigaTIME is a multimodal model that translates routine H&E (hematoxylin and eosin) pathology slides into virtual multiplex immunofluorescence (mIF) images across 21 protein channels — entirely from morphology, without any additional staining or re-sectioning of tissue. It was trained on a Providence dataset of 40 million cells and applied to 14,256 patients across 24 cancer types. The model operates at cellular resolution, inferring protein-specific fluorescence signals directly from brightfield pathology images.

GigaTIME unlocks the latent value of decades of archived H&E slides by enabling retrospective spatial proteomics at population scale, with direct implications for cancer diagnosis, prognosis, and treatment selection. Because H&E imaging is universal and cheap while physical multiplex staining is expensive and slow, this is a meaningful step toward equitable access to advanced pathology in resource-constrained settings. Within Microsoft Health Futures and the broader Providence collaboration, GigaTIME exemplifies how generative AI can extract new clinical information from existing data and accelerate precision-oncology research.