Unifying Standards and Ontologies to Enable Transparent, Context-Aware Imaging AI

Imaging informatics has grown far more complex than the systems originally designed to support it. Traditional PACS and RIS environments often lack the context needed to connect studies, findings, and patient history, leaving critical information siloed and workflows fragmented. The result is increased cognitive burden for radiologists, inefficiencies across teams, and barriers to personalization and advanced analytics. This session sets the stage by examining why today’s imaging ecosystem needs a shared, structured way to represent meaning—not just data—across systems and technologies. 

Attendees will learn how unifying standards and ontologies address these challenges by creating a common language that connects imaging concepts across modalities, vendors, and institutions. Participants will explore how a rigorously constructed radiology ontology enables structured, normalized, and interrelated representations of studies and series, making data comparable and searchable at scale. The session will show how this foundation reduces manual work, improves workflow intelligence, and allows clinical context—such as priors and longitudinal history—to be layered seamlessly into daily interpretation. 

Participants will also gain insight into how these same principles enable more transparent, trustworthy use of AI. Attendees will learn how standardized descriptors and metrics for AI models and datasets support reproducibility, fairness, and meaningful comparison, helping clinicians and organizations assess AI tools with confidence. By the end of the session, participants will understand how unified ontologies serve as essential infrastructure—powering context-aware AI, improving interoperability, and enabling more personalized, data-driven imaging care. 

Objectives

• Describe the limitations of traditional imaging systems and explain how unified ontologies reduce fragmentation and cognitive burden in radiology workflows.
• Explain how structured, normalized, and interrelated ontologies enable consistent classification and comparison of imaging studies across systems and vendors.
• Identify how standardized descriptors and metrics improve transparency, reproducibility, and trust in radiology AI models and datasets.
• Discuss how unified standards and ontologies support context-aware AI, interoperability, and more personalized, data-driven patient care.