Closing the gap in the clinical adoption of computational pathology

In recent years, a growing number of computational pathology methods have been developed to support diagnostic and prognostic tasks. These methods commonly employ artificial intelligence (AI) models, particularly deep learning (DL) architectures, to analyze digitized histopathological glass slides (whole-slide images) for applications such as tumor subtyping, prognosis prediction, and the inference of genetic alterations. However, the integration of such methods into routine clinical practice remains limited. In this presentation, I will first discuss the application of DL models in pathology. I will then present our proof-of-concept framework leveraging Health Level 7 (HL7) messaging and open-source resources for the seamless integration of AI-based models in the diagnostic workflow of a fully digitized pathology department (Angeloni et al., Genome Med. 2025).

• Understand the concepts of digital pathology and computational pathology

• Understand the challenges limiting the adoption of DL models in routine pathology diagnostics

• Understand the architecture and components of a standardized framework for deploying DL models in the diagnostic workflow of a fully digitized pathology department