Spaces:

---

raylim

/

mosaic-test

like 0

Sleeping

In cancer care, it’s very important not just to know that a tumor is “cancer,” but which kind of tumor it is, and what molecular features (mutations, biomarkers) it carries. Traditional approaches often lump together many detailed tumor types into broader classes, which can mask important differences in how tumors behave, respond to treatment, or relate to underlying biology. We built a machine-learning system that can look at tumor images (the classical stained slides used in pathology) and more precisely:


1. classify tumors into detailed subtypes,

2. infer their molecular features (e.g. what genetic changes or biomarkers they might carry), and

3. discover new relationships between tumor appearance (on the slide) and molecular traits.

https://pubmed.ncbi.nlm.nih.gov/40832221/