ACR Adopts First Practice Parameter for Imaging AI, Signaling a New Governance Era
The American College of Radiology has approved what it says is the first practice parameter for imaging AI, a notable move from experimentation toward formal clinical governance. The companion launch of the Assess-AI registry suggests the field is shifting from one-off validation studies to ongoing post-deployment monitoring.
The American College of Radiology’s approval of a first-ever practice parameter for imaging artificial intelligence is more than a professional-society milestone; it is a signal that AI in radiology is moving into the same governance framework that has long shaped imaging quality, safety, and reporting standards.
That matters because the core challenge in imaging AI is no longer whether the software can produce a promising result in a controlled study. The question is whether it can be used consistently across sites, scanners, populations, and workflows without quietly degrading performance or creating new failure modes.
The practice parameter also reflects a broader trend in healthcare AI: the industry is beginning to treat implementation as a clinical discipline rather than a procurement decision. By pairing standards with a registry, the ACR and SIIM are acknowledging that models need continuous oversight, not just premarket evidence.
If adopted widely, this could become a template for other specialties. Radiology has often been the first domain to formalize digital workflows, and AI may follow the same path from enthusiasm to normalization, with governance, auditing, and accountability becoming table stakes.