Advanced imaging in ankle fractures
Advanced imaging in ankle fractures
This dissertation of Jasper Prijs explores how advanced imaging and artificial intelligence (AI) can improve the diagnosis, understanding, and treatment of ankle fractures in both adults and children.
In Part I, the use of Convolutional Neural Networks (CNNs) is examined. Although CNNs show promising results in detecting and classifying fractures on X-rays and CT scans, external validation—crucial for clinical implementation—is often lacking. A newly developed model effectively detected, classified, and localized fibula fractures using a relatively small training dataset. Collaboration with Swedish researchers demonstrated the importance of international external validation. The clinical integration of AI is also discussed, emphasizing its role in decision-making and standardizing care.
Part II focuses on pediatric ankle fractures, specifically triplane fractures. Contrary to traditional beliefs, it was found that the configuration of these fractures is more influenced by ligamentous structures than by the degree of growth plate closure. Triplane fractures show strong similarities to adult trimalleolar fractures, suggesting a shared injury mechanism.
In Part III, subtle predictors of outcome following ankle fractures were investigated using postoperative CT scans and two-year Foot and Ankle Outcome Scores (FAOS). The presence of intra-articular loose bodies—regardless of size—was independently associated with poorer functional outcomes.
This dissertation highlights the added value of AI and advanced imaging in orthopaedic trauma care, while also emphasizing the need for robust model validation, trauma mechanisms, and attention to less visible factors that may influence recovery after ankle fractures.