TUTORIAL
From Pixels to Practice:
Clinical Pathology for AI Researchers
Level: Beginner to intermediate (no prior clinical or regulatory knowledge required)
Format: Mini-lectures and demonstrations
Prerequisites: None; All career stages welcome
September 27 | 08h35 - 10h35
AI models in computational pathology are growing more, yet most never reach the clinic. The gap is rarely algorithmic. Regulatory constraints, clinical accountability, fragmented infrastructures, and real-world workflow demands all shape whether a model ever sees routine use.
This tutorial bridges that gap. Jointly organized by COMPAYL and the European Society for Digital Pathology (ESDIP), it is designed for AI researchers working in digital or multimodal pathology and provides a clinically grounded perspective on how pathology is actually practiced, what translation into clinical workflows requires, and where AI support is most realistically impactful.
1. How pathologists think and work
Diagnostic reasoning, structured reporting, and lessons from real clinical environments led by a practicing pathologist.
2. Translational and regulatory realities
Non-technical barriers shaping model design, evaluation, and performance claims for researchers aiming beyond proof-of-concept.
3. Multimodal AI in pathology
Case-based tumor board walkthroughs integrating pathology, radiology, omics, and clinical data illustrating where AI adds genuine clinical value.
OVERVIEW
SPEAKERS
Haukeland University Hospital & University of Bergen, Norway
Practicing nephro- and gastrointestinal pathologist and researcher with expertise in structured pathology reporting, applied digital pathology, and diagnostic reasoning. Sabine leads the clinical component of the tutorial, guiding participants through real slide-based case discussions that illustrate how diagnoses are made and reported in routine practice.
Radboud University Medical Center,
The Netherlands
Senior Researcher and tenure-track faculty member specializing in multimodal AI for healthcare. She coordinates the tutorial and leads the multimodal integration session, facilitating tumor board–style cases that connect clinical decision-making, multimodal data integration, and the design and evaluation of AI systems.
Charité – Universitätsmedizin Berlin, Germany
Expert in digital pathology, AI translation, and interdisciplinary collaboration. Dr. Zerbe contributes to discussions on clinical validation and the practical realities of deploying AI in pathology workflows.
