Subject to change.
Subject to change.
Kenneth Philbrick PhD leads digital pathology ML externalization for Google Research’s Health AI Developer Foundations (HAI-DEF) program. Kenneth represents Google Research and Google Cloud at the Digital Pathology Association (DPA), DICOM Working Group 26, and has led Google's engagement at the digital pathology DICOM connectathons (2023 – Present). Kenneth is well published, over 45 peer reviewed manuscripts, in machine learning and the biological sciences (skeletal biology).
Over the past year, the Digital Pathology Association’s Regulatory and Standards Task Force has made measurable strides in shaping the regulatory and technical landscape of digital pathology. This session will highlight key initiatives and cross-functional collaborations that are accelerating the field's readiness for clinical and real-world deployment at scale.
Structured around the Task Force’s four core working groups—FDA Collaboration, AI/ML Oversight, Interoperability, and Best Practices—this 50-minute session will feature multiple presenters offering rapid-fire updates on achievements and future objectives. Topics will include how the DPA is working with the FDA to clarify regulatory pathways, evolving consensus on validating AI-enabled workflows, progress toward plug-and-play interoperability, and the development of practical best practice guidelines to guide adoption.
Opening remarks from Task Force leaders Dr. John Groth and Dr. Laura Chang will set the stage with reflections on the regulatory momentum gained over the past year and the shared goals driving this multi-stakeholder effort. Attendees will leave with an up-to-date understanding of where standards are headed—and how they can contribute to shaping the next wave of digital pathology innovation.
Learning Objectives:
1. Understand the Current Regulatory Landscape: Gain insight into how the Digital Pathology Association is collaborating with the FDA and other stakeholders to clarify and shape regulatory expectations for digital pathology platforms and AI-enabled tools.
2. Explore the Latest Standards Development Efforts: Learn about recent progress in interoperability frameworks, validation best practices, and data standards that support the scalable, reliable implementation of digital pathology in both clinical and research settings.
3. Identify Opportunities for Engagement and Adoption: Discover how you and your organization can contribute to or benefit from ongoing Task Force initiatives—whether by aligning with emerging best practices, participating in workgroup efforts, or preparing for regulatory trends.
DICOM is the interoperability standard through which digital pathology imaging is communicated. DICOM enables high performance image archival, metadata search and retrieval, and data / time efficient pixel imaging retrieval. Key for digital pathology, DICOM provides mechanisms to selectively retrieve sub regions of digital pathology images, to transcode imaging, and transform medical imaging to reference color profiles for interoperable display. In 2025, DICOM Working Group 26 conducted a Connectathon to test end-to-end interoperability, scanner-to-archive-to-consumer(viewer or ML), between vendors implementing actors using the DICOM and IHE defined interoperability standards. The 2025 connectathon brought together the largest collection of vendor participants, N=30 vendor participants, spread across image acquisition, archival, viewer, and ml actor rolls. The connectathon was a resounding success demonstrating the ability of multiple slide scanner vendors to generate and store medical imaging with LIS metadata within multiple DICOM VNAs. Viewing software and ML systems were able to retrieve the imaging and generate annotations which were themselves stored as DICOM within the participating archives. These results demonstrate that systems built to support the existing interoperability standards can be easily connected to implement end-to-end enterprise imaging workflows. The DICOM standard is an evolving standard that grows in response to vendor / customer technical advancement. Current areas of active work include, the recently added support for JPEGXL encoded imaging, and defining the standards representation for multi-spectral imaging, imaging with multiple focal points (z-stacks), and vector graphics annotation with holes.