The Imaging Problem List: A Standards-Based Framework for Longitudinal Tracking of Imaging Findings

Preprint. This manuscript is under peer review. Objective. Imaging findings are longitudinal clinical entities, yet radiology reports treat each observation independently. Existing standards provide structured representation for individual findings but do not track them over time. We propose the Imaging Problem List (IPL), a standards-based framework that aggregates structured imaging observations into persistent entities tracked across time and modalities. Methods. We designed the IPL entry as a persistent record that aggregates observations of the same finding into a longitudinal history, with graduated-confidence linking proposed to connect findings across reports. The framework builds on standards-coded imaging observations and is designed to be populated by foundation models that generate and update entries from current and prior reports, and eventually the images, under radiologist verification. Results. We demonstrated the framework using a synthetic patient with chronic liver disease tracked across four modalities over seven years. Temporal aggregation grouped 223 observations across 15 examinations into 41 IPL entries (5.4:1). Findings included progressive measurement increase approaching a surgical threshold, reclassification after documented stability, acute-to-chronic evolution, and resolution with residual sequelae. Discussion. The IPL gives radiology a longitudinal memory it lacks: a structured substrate artificial intelligence (AI) can both generate and reason over. Applications could include pre-population of chronic findings, structured follow-up tracking, and cross-subspecialty finding awareness. Conclusion. The IPL extends structured imaging standards into the temporal dimension, giving imaging findings the persistence to be tracked, trended, and acted upon as laboratory values and diagnoses long have been. It also gives AI a structured foundation to generate and maintain that record.