Radiologists' perspectives on the use of artificial intelligence in emergency radiology: A pilot survey

Artificial intelligence (AI) is increasingly being explored in radiology, including its potential role in emergency imaging settings. However, global perspectives on AI adoption, usefulness, and limitations among emergency radiologists remain underexplored.

To assess awareness, usage, perceived benefits, and limitations of AI tools among radiologists practicing emergency radiology worldwide.

A 16-question survey was distributed globally between October 24, 2024, and August 4, 2025, targeting radiologists working in academic, community, and private settings who practice emergency radiology as a primary or secondary subspecialty. The survey was disseminated via direct emails extracted using automated and manual methods from recent publications in major radiology journals. A total of 57 responses were collected.

AI awareness was high (93%), but frequent clinical use was reported by only 28%. Daily use of AI in emergent imaging was limited to 23% of respondents. The majority anticipated AI becoming essential within five years (68%), and 51% believed AI would replace certain radiological tasks. Image interpretation and acquisition were the most common AI applications. Key perceived benefits included improved diagnostic accuracy and increased efficiency, while concerns included limited accuracy, integration difficulties, and cost. Trust in AI varied by experience, with less experienced radiologists viewed as more trusting.

While emergency radiologists globally recognize AI's potential, significant barriers to its routine adoption remain. Addressing issues of trust, cost, accuracy, and workflow integration is essential to unlock AI's full utility in emergency radiology.