Summary

Virtual microscopy has been relied upon in the research setting for years, but the adoption of digital pathology in clinical practice has stalled. In contrast to radiology, where digital workflows simplified hospital operations and reduced costs, digitizing tissue sections in pathology adds steps to laboratory workflows without eliminating the glass slides. Additionally, the costs for IT infrastructure are significant. The value proposition of digital pathology currently hinges on artificial intelligence, which promises to unlock its full potential. However, making digital and computational pathology a clinical reality is challenging due to the risk and complexity of histopathological diagnostics and poses significant technical, operational, and regulatory hurdles that are difficult to overcome by hospitals alone given the interdisciplinary nature of the task and the interdependency of stakeholders.

Objectives

• Describe the current digital pathology landscape

• Discuss the value proposition for digital pathology, especially as it contrasts with radiology

• Explore strategies for de-identifying and sharing pathology and radiology image data with the research community to enable machine learning projects that require large amounts of data

• Explain how to curate data sets and design digital reader studies for artificial intelligence applications with a regulatory science focus