| Abstrakt: |
A recent study conducted at the Peter MacCallum Cancer Centre in Victoria, Australia, explored the use of artificial intelligence (AI) models for auto-segmentation in pediatric computed tomography (CT) data sets. The study found that AI models trained exclusively on adult data performed poorly on pediatric data, particularly for children aged 0 to 2. However, the addition of pediatric training data significantly improved the performance of the models for all age groups. The study also found that the AI models demonstrated robust cross-scanner generalization, suggesting their potential for real-world clinical use. The researchers emphasized the importance of including diverse data sets in training AI systems for medical image interpretation tasks. [Extracted from the article] |
