The ESR, RSNA, and three other societies have issued a joint statement on the development and use of AI in radiology.
The statement is a collaborative effort between the ESR, RSNA, American College of Radiology (ACR), Canadian Association of Radiologists (CAR), and the Royal Australian and New Zealand College of Radiologists (RANZCR). It was written by a team led by Prof. Adrian Brady, chair of the ESR Board of Directors, and published on 22 January in the RSNA journal, Radiology: Artificial Intelligence.
The paper "defines the potential practical problems and ethical issues surrounding the incorporation of AI into radiology practice," and delineates "the main points of concern that developers, regulators, and purchasers of AI tools should consider prior to their introduction into clinical practice" and offers "methods to monitor the tools for stability and safety in clinical use, and to assess their suitability for possible autonomous function."
"This statement will serve as both a guide for practicing radiologists on how to safely and effectively implement and use the AI that's available today, and a roadmap for developers and regulators on how to approach delivering improved AI for tomorrow," noted co-author Dr. John Mongan, PhD, of the University of California, San Francisco and chair of the RSNA Artificial Intelligence Committee.
![A normal mammogram confirmed by three-year radiologic follow-up illustrates reader-marked regions of interest (ROIs) during (A) unaided (round 1) and (B) artificial intelligence (AI)–assisted (round 2) reading. Each colored dot represents an ROI for recall by a human reader. Readers could mark more than one ROI per case, represented by multiple dots of the same color. During AI-assisted reading, the AI system displayed three visible prompts: two with suspicion of malignancy scores of 35% (left mediolateral oblique [L MLO] and craniocaudal [L CC]) and one with a suspicion of malignancy score of 10% (right craniocaudal [R CC]), shown as polygonal overlays. Without AI, six of 10 readers (60%) marked a false-positive ROI. With AI assistance, this fell to two of 10 (20%). R MLO = right mediolateral oblique.](https://img.auntminnieeurope.com/mindful/smg/workspaces/default/uploads/2026/07/2026-07-14-radiology-mammogram-ai-auto-bias.H0bYO8QlWs.jpg?auto=format%2Ccompress&fit=crop&h=100&q=70&w=100)
![A normal mammogram confirmed by three-year radiologic follow-up illustrates reader-marked regions of interest (ROIs) during (A) unaided (round 1) and (B) artificial intelligence (AI)–assisted (round 2) reading. Each colored dot represents an ROI for recall by a human reader. Readers could mark more than one ROI per case, represented by multiple dots of the same color. During AI-assisted reading, the AI system displayed three visible prompts: two with suspicion of malignancy scores of 35% (left mediolateral oblique [L MLO] and craniocaudal [L CC]) and one with a suspicion of malignancy score of 10% (right craniocaudal [R CC]), shown as polygonal overlays. Without AI, six of 10 readers (60%) marked a false-positive ROI. With AI assistance, this fell to two of 10 (20%). R MLO = right mediolateral oblique.](https://img.auntminnieeurope.com/mindful/smg/workspaces/default/uploads/2026/07/2026-07-14-radiology-mammogram-ai-auto-bias.H0bYO8QlWs.jpg?auto=format%2Ccompress&dpr=2&fit=crop&h=167&q=70&w=250)
