AI in Pathology
Pathology AI depends on whole-slide scanning, so it is strongest where labs are digital. Tools for prostate biopsy grading and cancer detection are in use in some labs. Foundation models trained on millions of slide tiles are the main research direction.
What AI does in Pathology
Marks suspicious areas and suggests Gleason grade for the pathologist to confirm.
Examples: Paige Prostate
Finds small tumour deposits in sentinel nodes, a task tested in the CAMELYON challenges.
Examples: CAMELYON-trained algorithms
Counts Ki-67, HER2 or PD-L1 positive cells more consistently than eyeballing.
Examples: QuPath (open-source), commercial image analysis
Flags out-of-focus slides and sorts cases by likely urgency.
Examples: Workflow tools from slide-scanner vendors
Predicts molecular features such as MSI status from routine slides. Used to prioritise testing, not replace it.
Examples: UNI, CONCH, Virchow
Search a slide archive with text such as signet ring cells.
Examples: CONCH
routine = in everyday use in many centres · early adoption = in use at some sites, evidence still building · research = studies only
Key ideas for this specialty
New to the basics? Start with the theory lessons.
Tools
Open-source weakly supervised slide classification code from the Mahmood lab.
Regulatory status: Not listed
Open pathology foundation model from the Mahmood lab (access terms apply).
Regulatory status: Not listed
Lymph node metastasis dataset and benchmark.
Regulatory status: Not listed
Commercial pathology AI company with prostate and other cancer tools.
Regulatory status: Not listed
A listing here is not an endorsement. Ask any vendor for its CDSCO licence and class before clinical use.
Where to learn
Learn slide analysis on a free tool.
Broad review by Song and colleagues.
Explains tasks from detection to outcome prediction.
Good first read on the field.
General medical AI grounding.
Do it yourself
Download QuPath, open a public whole-slide image, and run the built-in cell detection. Compare the counts with a manual count on one region.
Open the dataset or tool →Read the CAMELYON17 task page and view example slides with tumour annotations. Write down how many cases the training set holds and how labels were made.
Open the dataset or tool →Use public or properly de-identified data only, and get ethics approval before any project on your own patients.
Papers worth reading
- Towards a general-purpose foundation model for computational pathology Nature Medicine, 2024
- A visual-language foundation model for computational pathology Nature Medicine, 2024
- A foundation model for clinical-grade computational pathology and rare cancers detection Nature Medicine, 2024
- Artificial intelligence for digital and computational pathology Nature Reviews Bioengineering, 2023
Societies and guidance
Source of open pathology foundation models; a research lab, not a society.
Open challenge platform used by pathology AI groups.
Recent updates
FICCI-Praxis report names AI diagnostics as the first large-scale medtech use in India
The report 'AI in MedTech' was released by Union Health Minister J.P. Nadda at India Medical Device 2026. It lists data readiness, regulation and reimbursement as the three gaps between pilots and wide use.
Source: Medical DialoguesGoogle releases MedGemma 1.5 and MedASR as open models
MedGemma 1.5 4B reads CT, MRI and histopathology images and runs locally, and MedASR is a medical dictation speech model that Google says makes 58% fewer errors on chest X-ray dictation than general models.
Source: Google ResearchLinks checked 3 October 2026. “Not listed” means we could not confirm the detail, not that it is absent. Educational material, not clinical advice.