OnCo
technologiesTechnologyEstablished

Digital pathology & AI

Scanning microscope slides and letting software measure things a pathologist cannot see, including predictions of who will benefit from a treatment.

Whole-slide imaging is now routine in many centres. FDA-cleared tools: Paige Prostate (detection), ArteraAI Prostate (2025, first prognostic and predictive AI) and ArteraAI Breast (May 2026, risk stratification in early HR+/HER2- breast cancer). Foundation models (Virchow, UNI, CONCH, Prov-GigaPath) predict molecular status (MSI, HRD, HER2) from H&E alone.

Schematic · not to scale
Stained protein (HER2, PD-L1) · Glass slide · Whole-slide scan → model

How it works

Gigapixel whole-slide images; tile-level self-supervised encoders aggregated to slide-level predictions.

Strengths
  • Cheap biomarker from routine slides
  • Consistent scoring (Ki-67, TILs, HER2)
Limitations
  • Scanner and stain domain shift
  • Explainability
  • Regulatory pathways for updates

Products

2top

Latest papers

top
Literature trend424 papers in the last 12 months+67% vs prior 12How this is computed
Latest papers · live from Europe PMC
Open in Europe PMC

Query for this technology: (TITLE:"digital pathology" OR ABSTRACT:"digital pathology" OR TITLE:"computational pathology" OR ABSTRACT:"computational pathology" OR TITLE:"whole slide image" OR ABSTRACT:"whole slide image") AND (deep learning OR artificial intelligence). Results are unfiltered search hits about Digital pathology & AI, not a curated reading list.

Connected

72top

Pages like this

not linked directly; found by shared links

cancers

3

fronts

2

technologies

5

drugs

2

companies

7

institutions

6

terms

6

pairings

1

roadmaps

1

ideas

33
A dedicated fund for randomised trials of cancer AI with patient outcomesA federated learning consortium of cancer centres that jointly own the modelsA fund for prospective validation of academic biomarkers and companion diagnosticsA pre-competitive consortium to train a shared multimodal cancer foundation modelA registry of external validation datasets for cancer AI models, with mandatory reportingA same-week expert second opinion for every rare cancer diagnosisA standard evaluation pathway for AI-assisted pathology, from reader study to deploymentA video-based surgical quality registry linking assessed skill to cancer outcomesAI quantification of HER2-low and HER2-ultralowAI second reads to stop borderline lesions being upgraded to cancerAI-first reading for high-volume common cancer diagnoses, pathologist for the exceptionsCalibrated reference slides so every lab scores HER2-low the same wayClear the suppressive neutrophils out of pancreatic tumours firstContinuous prospective validation for every oncology AI tool after deploymentDigitise the nation's pathology slides and link them to outcomesFederated training of pathology and radiology models across hospitalsMap which tumour clones sit next to which immune cells before choosing therapyMatch therapy to the type of scar-forming cell in the tumourMobile diagnostic units that biopsy, scan and treat on the same visitOne certified open-source de-identification pipeline for scans and slidesOne digital PD-L1 scale that maps across all the competing assaysPathologist assistants plus AI triage to multiply pathologist capacityPatient-level multimodal foundation models for treatment selectionPublic gold-standard datasets for validating every cancer biomarker testReal-time margin assessment and image-guided surgery as the global standardSequestered, prospectively collected benchmark datasets that no one can train onSlide scanners in district hospitals wired to pathologists anywhereStandards for spatial and multiplex tissue biomarkers before they reach the clinicThe pathology lab triggers a trial referral the day a rare cancer is diagnosedTIL-based omission of chemotherapy in stage I TNBCTurn the map of immune cells inside a tumour into a standardised testVersion control and locked reference sets for AI algorithms used as companion diagnosticsWhole-patient digital twins validated in prospective randomised trials

bottlenecks

6