2024 USCAP Seminar





Monday, March 25, 12-1 PM Eastern

Baltimore Convention Center, Seminar Room 5 (Exhibit Hall)


The Data, Regulatory, and Ethics of Augmented Intelligence (AI) in Pathology


With the continuous advancement of AI in pathology, the importance of quality data set, regulatory, and ethics become practical issues that worth careful consideration.  Experts from the academics, FDA, and industry will share the principles and best practices of collaborative data consortium, the literature review and update on thoughts regarding AI ethics.  A Q&A session will follow presentations from our thought leaders. Attendees are welcome to meet Digital Pathology Association leadership to explore networking and professional development opportunities.


Learning Objectives:

  • Discuss various examples of accessing quality pathology image data for AI development.
  • Highlight an NCI funded CHTN centers to create a Federated model for secure sharing of patients imaging data (radiology and pathology) in a HIPAA compliant and ethical manner.
  • Discuss all things considered in AI ethics.
  • Spotlight the accomplishments of DPA to encourage participation and collaboration.


DPA Highlights by Michael Quick

The remarkable achievements of the Digital Pathology Association (DPA) will be reviewed. Formed to advance the field of digital pathology, DPA is driving improvements in modern medicine. Through promoting best practices, educational resources, and research collaboration, DPA is bridging the gap between technology and pathology. DPA’s commitment to advocating for standards has fostered global collaboration. Delving into DPA's milestones, it becomes evident of its pivotal role in merging traditional pathology with cutting-edge technology, ensuring quicker, accurate diagnoses and improved patient outcomes.


Collecting and Annotating Digital Pathology Images to Assess Computational Pathology by Brandon Gallas

This presentation will summarize the efforts to create a dataset of pathologist annotations of stromal tumor-infiltrating lymphocytes on whole slide images of triple negative breast cancer biopsies. The dataset’s purpose is to assess the performance of computational pathology models. The lessons learned during the pilot study, the pathologist training (a CME course and interactive training modules), and the pivotal study status will be highlighted. A framework for reporting the collection of annotations to permit transparency and reproducibility and initial discussions with FDA reviewers on a submission to the medical device development tool program regarding the dataset will be discussed.


Building of effective and quality pathology image data collection for AI development by Rajendra Singh

Diversity of data is the most critical requirement for building and validating AI models. Diverse datasets can come from multi-institutional supported On Demand Biorepositories. On Demand Biorepositories works prospectively with each investigator to tailor specimen acquisition and processing to meet their specific project requirements and provide access to the data in a secure and compliant ecosystem. In this presentation we will present the mechanism for on demand biorepositories to share data as well as be HIPAA Compliant and follow all ethical principles of data sharing.


Large Language Models in Pathology– Ethical Considerations of an Emerging AI Paradigm by Rama Gullapalli

Large Language Models (LLMs) are a new and powerful paradigm in the field of artificial intelligence (AI). Though the technology is only a few years old, there are dramatic examples of LLM use in different specialties including health care and pathology. Initial case studies of LLM usage were focused mainly on the domain of text and natural language processing. However, LLMs usage has now expanded to include multi-modal data types including image data. Naturally, pathology is likely to be one of the major targets for the use of these enhanced LLMs (known as “foundational models”) within healthcare. While powerful, there remain many unanswered questions. In the current presentation, we will discuss some of these key issues that a practicing pathologist must be aware of. An overview of topics such as biases, hallucinations, distributive shifts, and chronic model behavior associated with LLMs will be discussed. The promise of LLMs (and foundational models) within healthcare is bright. However, pathologists must be cautious in adopting this technology.  Professional organizations (such as USCAP and CAP) have a critical role in the adoption of AI-LLM technologies by establishing the proper guidelines of usage. AI technologies like LLMs are immensely powerful, and when harnessed appropriately, have the potential to dramatically change the way pathology is practiced in the future.




Bui_newMarilyn M. Bui, MD, PhD

DPA Past President

Senior Member and Professor of Pathology

Scientific Director of Analytic Microscopy Core

Moffitt Cancer Center & Research Institute








Brandon D. Gallas, PhD 


Division of Imaging, Diagnostics, and Software Reliability




Rama GullapalliRama Gullapalli, MD, PhD

Assistant Professor

Department of Pathology, Chemical and Biological Engineering

University of New Mexico Health Sciences Center




Mike QuickMichael Quick

2024 DPA president

Vice President, R&D, Oncology, Cytology




Raj HeadshotRajendra Singh, MD

Director of Digital Pathology and Dermatopathology

Summit Health





Bethany WilliamsBethany Williams MBBS, BSc, PhD 

Lead for Digital Pathology Education and Training

National Pathology Imaging Co-Operative







This Exhibitor Seminar is not a part of the official USCAP Educational Program at the 2024 Annual Meeting, and is not sponsored by the USCAP.  The USCAP does not officially endorse any company or its products and does not award CME credits for attendance at Exhibitor Seminars.