A recent international survey of paediatric radiology leaders, published in Eur Radiol, indicates that while artificial intelligence (AI) is seeing growing adoption in clinical settings, its overall impact on patient care remains modest. Researchers found that 88.9% of the 18 surveyed global institutions have already implemented at least one AI application, yet many leaders cited significant systemic barriers to further progress.
Bone age assessment emerged as the most frequent use case, accounting for 44.4% of implementations, followed by image segmentation, protocol optimization, and natural language processing. Despite these deployments, the average clinical impact of these tools was rated at 3.56 out of 5, with only 16.7% of respondents characterizing the effect as truly transformational.
The study highlights a critical dichotomy: while technical development of algorithms continues, the primary obstacles to success are organizational and data-related. A substantial 83.3% of participants identified the scarcity of paediatric-specific datasets as the leading challenge. Other significant hurdles include integration with existing clinical workflows (66.7%), high costs, and cybersecurity concerns.
Experts suggest that the focus of the field must shift. More than half of the respondents (55.6%) agreed that current research is overly concentrated on model development at the expense of clinical integration, ethics, and long-term sustainability. The findings emphasize that future advancements will likely depend on multicentre collaboration to resolve data gaps and the presence of internal institutional champions to bridge the gap between technology and daily clinical practice.