GenVinci builds 3D tissue models from flat microscope images
The computational model predicted gene activity from tissue images and mapped states linked to senescence across 85.8 million cells.

Our read
AI & data- Studied:
- Preprint: computational study of tissue images from mouse organs and human skin
- What changed:
- GenVinci reconstructed three-dimensional tissue maps from two-dimensional microscope images
- The catch:
- Reconstructed gene-activity maps substitute for direct measurements throughout intact three-dimensional tissue
Editor of record: Boris Djordjevic · AgingResearch.News
Aging mouse organs and human skin spanning five decades supplied 1,782 tissue sections for a resource used to develop GenVinci, a computational model. The wider resource contained 300 million measured or predicted profiles of gene activity and 4,623 tissue sections. GenVinci reconstructed three-dimensional tissue maps from routine two-dimensional microscope images and predicted gene activity across tissue.
The model mapped states associated with senescence, a state in which cells stop dividing, across 85.8 million cells. It revealed patterns specific to different cell types and changes associated with disease. The authors presented GenVinci as a framework for mapping senescence across different scales.
Why it matters
The framework offers a way to study how senescence-associated cell states fit into the three-dimensional organization of aging and diseased tissues.
Caveats
The maps are computational reconstructions, rather than direct measurements of gene activity throughout intact three-dimensional tissue. The work is a preprint and has not been peer-reviewed.
- Foundation models
- Resources: datasets, atlases and tools
- Cellular senescence
- Atopic dermatitis
- Keratinocyte cancer
- Humans
The paper
Massachusetts General Hospital; Broad Institute
bioRxiv, Nov 18, 2026, Open access, CC BY-NC-ND, Preprint, not yet peer-reviewed



Comments
Be the first to comment.