The agenda for the Digital Pathology & AI Europe conference has been posted over at Linked Iin by Lauren Dennison. Dates are 9-10 December in London. See the 15 page PDF agenda. Novotel London West (just west of Kensington).
# #
AI CORNER
# #
London’s Digital Pathology Congress Shifts From Adoption to Full-Scale Operations
The 13th Digital Pathology & AI Congress: Europe will take place in London on December 9–10, 2026, bringing together more than 500 attendees and over 60 speakers from pathology laboratories, health systems, pharmaceutical companies, universities, regulators, and technology vendors. The organizers’ central message is clear: digital pathology is moving beyond the question of whether laboratories should digitize. The more urgent questions now concern how to make digital operations reliable, interoperable, scalable, and ready for routine artificial intelligence.
Three tracks
- The meeting is organized around three parallel tracks.
- The first concentrates on implementation, workflow, and standardization.
- The second covers image analysis, computational pathology, and AI.
- The third follows the growing use of digital pathology and AI in pharmaceutical research, moving from preclinical applications on the first day to clinical development and companion diagnostics on the second.
Practical Realities
A major emphasis will be the practical realities of operating digital pathology across large, multi-site systems. Speakers from the National Pathology Imaging Co-operative, the Wessex Pathology Network, NHS Wales, Liverpool, Southampton, and other organizations will address scanner deployment, laboratory workflow redesign, image management, data storage, training, governance, quality assurance, and user acceptance. These sessions should be particularly valuable for organizations that have purchased scanners but discovered that digitization is not simply a matter of replacing glass slides with electronic images.
Interoperability!
Interoperability appears repeatedly throughout the program. Presentations will examine how laboratories can connect scanners, laboratory information systems, image-management platforms, AI applications, and enterprise data systems without becoming trapped in disconnected vendor environments. Yale’s Peter Gershkovich will describe an open, modular pathology ecosystem, while Source LDPath and the European Institute of Oncology will offer practical lessons from integrating heterogeneous scanners, LIMS platforms, and hospital systems. The organizers clearly view architecture and workflow integration as prerequisites—not afterthoughts—for successful AI deployment.
Pragmatic AI?
AI itself will be approached with a noticeably pragmatic tone. Paul van Diest’s opening keynote, “Use of AI in Diagnostic Pathology: Just do it?,” asks whether regulatory authorization alone is enough to justify routine use. Other sessions will examine real-world AI deployment in breast pathology, dermatopathology, biomarker scoring, veterinary pathology, toxicologic pathology, and laboratory quality assurance. The program extends well beyond familiar tumor-detection algorithms to include foundation models, agentic systems, prediction of RNA-associated biology from tissue images, spatial biomarkers, virtual staining, and laser-capture microdissection.
Whither Regulation?
Regulation and validation will receive substantial attention. Representatives of the UK Medicines and Healthcare products Regulatory Agency will report on the AI Airlock regulatory sandbox, including lessons from digital pathology products tested against real-world regulatory challenges. A dedicated panel will address quality assurance, while another will focus on the “innovation gap” between promising clinical AI models and successful implementation. Pharma sessions will likewise emphasize fit-for-purpose validation, lifecycle governance, data quality, revalidation triggers, and the evidence regulators require.
R&D Pharma
Drug developers will find a particularly strong program. Speakers from AstraZeneca, GSK, Roche, Bayer, Novo Nordisk, MSD, Immunocore, Charles River Laboratories, Natera, and others will discuss toxicologic pathology, virtual staining, clinical-trial workflows, AI-enabled biomarkers, minimal residual disease, spatial transcriptomics, and companion diagnostics. A panel on companion diagnostics in the AI era will consider predictive pathology, spatial biology, oncology co-development, and the validation of AI biomarkers alongside therapeutics.
Who should go?
The congress should appeal to several overlapping groups: pathology departments preparing for enterprise digitization; laboratories struggling with integration or workflow redesign; pathologists evaluating AI tools for routine use; informatics and laboratory leaders responsible for data architecture; regulators and quality specialists; pharmaceutical pathology and translational-medicine teams; and AI developers seeking a better understanding of clinical evidence requirements. Technology vendors will also be present in force, with major scanner, software, image-management, and AI companies represented among the sponsors and exhibitors.
Overall, the London meeting is designed less as an introduction to digital pathology than as a working conference for a field entering its operational phase. Its most important themes are scale, interoperability, validation, governance, and the conversion of increasingly sophisticated AI research into dependable clinical and pharmaceutical workflows.
Sidebar: Five Unexpected Highlights in London
1. AI may infer RNA biology directly from tissue images.
One of the more futuristic sessions will examine whether self-supervised models can recover RNA-associated biological programs from ordinary pathology and microscopy images. The work spans breast, ovarian and pediatric kidney cancers—and even neurodegenerative disease—suggesting that morphology may contain molecular information that conventional visual review cannot readily extract.
2. Alpha-particle physics could redefine spatial biomarkers.
A Bayer presentation will argue that average biomarker expression may be the wrong measurement for targeted alpha therapies. Because alpha particles travel only about 45–100 micrometers in tissue, therapeutic effect may depend on where target-positive cells sit relative to that fixed “kill radius.” The proposal is to anchor spatial pathology measurements to physics rather than pathologist impression alone.
3. Veterinary pathology is becoming an AI proving ground.
The Royal Veterinary College will describe transfer-learning models applied to animal tumors and other diseases, combining whole-slide images with immunohistochemistry, molecular diagnostics, imaging and surgical data. Veterinary pathology may offer an unusually rich setting for discovering hidden morphological patterns and developing tools that could later inform human medicine.
4. “Virtual staining” could reduce the need for actual stains.
GSK will present work on generating stain-like images from unstained or minimally processed tissue. The potential advantages include faster workflows, less reagent use and preservation of scarce tissue. But the session also emphasizes the unresolved issues: morphological fidelity, biological validity, reproducibility and defining which uses are safe enough for research or eventual clinical adoption.
5. Pathology AI is becoming “agentic,” not merely predictive.
A University of Lausanne presentation will move beyond algorithms that classify a patch or score a slide. The proposed next generation of systems could reason across whole-slide images, retrieve molecular and clinical context, and coordinate multiple models and tools within a diagnostic workflow. That is a significant conceptual leap—from AI as a specialized image reader to AI as an active participant in the pathology information system.
