Application of Artificial Intelligence in Paediatric Imaging

Artificial intelligence (AI) has been intensively explored in various fields of medical imaging, such as neuroradiology and oncological imaging. Comparatively, applying AI in paediatric imaging has received insufficient attention. Research on paediatric imaging techniques and clinical diagnosis is significantly lagging behind that of adults for several reasons. First, paediatric medical imaging data are much less abundant than their adult counterparts, leading to a lack of large-high-quality databases for AI system development. Second, obtaining paediatric medical imaging data and ensuring data homogeneity are more challenging than in adults. For instance, children often have difficulty complying with paediatric imaging examinations, and sedation may be needed to ensure that the procedure is completed successfully. In addition, children at different stages of physiological development may result in significant differences in normal data. Third, there is a significant overlap between the pathological and physiological changes in paediatric diseases and the differences in healthy child growth and development, which affects data uniformity and quality. As a result, children are excluded from healthcare innovations driven by AI designed for adults; however, applying adult-oriented AI software in paediatric care may pose potential risks [1].