Dear AuntMinnie Europe Member,
This issue repeatedly returns to the chest x-ray, unsurprisingly. One of radiology’s oldest and most widely used examinations has become a testing ground for some of its newest challenges: optimizing acquisition, reducing low-value imaging, recognizing artifacts, and ensuring AI performs fairly and improves patient care.
In this edition, we examine an Irish-led review showing that more than two decades of research have failed to establish an optimal beam quality for adult chest radiography. After analyzing 29 studies, the researchers called for standardized methods and reporting to support safer, more consistent protocols.
We also explore a fairness-aware pneumonia AI model developed by German researchers. The strategy reduced several sex-based performance disparities without compromising overall accuracy, but widened the sensitivity gap, demonstrating that improving one fairness measure can shift inequalities elsewhere.
Another study tested a commercially available fracture-detection system using 1,500 consecutive cases from hospitals in Denmark, Germany, and the Netherlands. The device performed consistently across the three centers and improved clinician sensitivity by 11 percentage points, although results varied for certain anatomical subgroups.
In pediatric imaging, researchers found that 73% of chest x-rays performed in preschool children with recurrent wheezing showed abnormalities. Yet only 1% prompted further investigation, and none changed the diagnosis or treatment, supporting selective rather than routine imaging in this population.
Researchers presenting at UKIO 2026 also showed how tattoo pigments can mimic pathology and introduce artifacts on x-rays. Yellow and white pigments were the most detectable, while the findings underline the importance of knowing when features outside the suspected disease process may affect image interpretation.
Finally, a real-world implementation study from Spain found that AI-assisted chest x-ray triage was associated with a 24% relative reduction in pulmonology referrals without increasing subsequent CT use. The findings suggest that AI may deliver measurable clinical value when integrated into a radiologist-supervised workflow.
For more x-ray news, be sure to check in regularly with our Digital X-Ray content area. As always, if you have x-ray topics you would like us to consider, please drop us a line.
Claudia Tschabuschnig
Associate Editor
AuntMinnieEurope.com