Just had the pleasure of being interviewed by
@Nature on the rise of AI in digital pathology. With growing workloads and global shortages of pathologists, the field is turning to AI not as a luxury—but as a necessity.
Full article:
nature.com/articles/d41586-0…
In the article, I discuss how recent advances in foundation models like UNI-2 and CONCH are redefining what’s possible in cancer diagnostics. Trained on hundreds of millions of pathology patches, these models go beyond classification: they enable molecular subtyping, caption generation, and even zero-shot inference.
But while the hype is real, so are the challenges. Cross-site generalization, lack of external validation, and regulatory hurdles remain major barriers. We must invest in robust benchmarking, multi-institutional trials, and trustworthy model design to ensure AI truly supports—not replaces—clinical judgment.
Digital pathology isn’t the future—it’s already here. Let’s make it safe, scalable, and equitable.
@UHN_Research @PMResearch_UHN @VectorInst @UofT