Artificial Intelligence in Patellar Pathology: A Narrative Literature Review

Disorders of the patellofemoral joint (PFJ) are a common source of knee pain across all age groups and can impact mobility and quality of life. These disorders, such as patellar instability, anterior knee pain, and patellofemoral joint osteoarthritis (PFOA) are not only frequent, but often persist or recur even after treatment. This impairs the ability of many patients to return to their previous level of activity. A growing body of research is now looking into how anatomic features like patellar height and shape contribute to these pathologies. More recently, artificial intelligence (AI) and machine learning (ML) have been used, offering new tools for evaluating knee pathology, particularly through medical imaging. While there's already promising evidence for AI in diagnosing and predicting tibiofemoral osteoarthritis, its use in patellar pathology is still developing. This review aims to summarize and evaluate existing studies that have applied AI to aspects of patellar pathology–especially in diagnosis, monitoring progression, and making predictions. We conducted a structured review of relevant studies up to December 2024, focusing on those that used AI/ML models in the assessment of PFJ conditions based on imaging, with a specific emphasis on patellar shape, height, cartilage damage, and joint degeneration. Out of 126 initially identified studies, 16 met the criteria for inclusion. These papers explored AI's utility in diagnosing PFOA, segmenting the PFJ, measuring patellar height, analyzing cartilage, and predicting patellar instability. Findings show that several AI models, including convolutional neural networks (CNNs), YOLO, ResNet, and U-Net architectures, have demonstrated strong performance. They often matched or exceeded traditional radiologic assessment. Combined models that integrate clinical data with imaging tend to perform best, especially for PFOA diagnosis and progression prediction. In addition, AI-based measurements of patellar height and geometry displayed strong reliability in predicting instability in both pediatric and adult populations. Overall, AI continues to emerge as a valuable tool for improving how we understand, diagnose, and monitor patellar conditions, with the potential to improve clinical decision-making and patient outcomes. Further research is warranted to refine these tools and validate their use in real-world orthopedic settings.