Artificial Intelligence in Pediatric Imaging: A Primer for Pediatric Clinicians

Artificial intelligence (AI) is reshaping healthcare, and radiology is at the forefront for adoption. Increasing demand for imaging, complex protocols, quantitative image analysis, and workforce shortages drive AI development. However, clinical translation in pediatric imaging still lags its use in general radiology. Only a small number of approved tools carry explicit pediatric indications, and these focus on narrow tasks. In chest imaging, deep learning models can detect pneumonia, tuberculosis, and misplaced lines or tubes, though performance depends on pediatric-specific training and local validation. Automated bone age assessment is among the earliest and most widely adopted applications. Neuroimaging benefits include reduced acquisition times, lower contrast doses, and rapid triage of critical findings. Cardiovascular applications support congenital heart disease detection and functional assessment, while oncology applications are being explored for tumor characterization and segmentation for therapy planning. Beyond interpretation, AI can enhance workflow triage, structured reporting, and patient communication. Despite these advances, barriers remain for AI adoption. These include scarcity of pediatric datasets, limited prospective validation, workflow integration challenges, interpretability concerns, and complex ethical and legal issues unique to children. Integration with hospital systems is uneven, and routine monitoring and support are often missing. Pediatricians and surgeons should work with radiologists and data scientists to define use cases, curate datasets, validate performance, and set ethical guardrails. Future progress will depend on collaborative data sharing, robust validation, clinician engagement, child-centred guidelines, and human-centred design to ensure safe, equitable, and meaningful adoption.