Imaging of Wrist Ligament Pathology: Current Status and Emerging Role of AI

Abstract Wrist ligament injuries are common sources of pain, instability, and functional impairment, but their diagnosis remains challenging. Several imaging modalities are used in clinical practice, including radiography, ultrasound, computed tomography, and magnetic resonance imaging (MRI). Each provides useful information, but all have important limitations. Small ligamentous structures may be difficult to visualize directly, partial tears may be missed, and dynamic instability may remain occult on static imaging. Among these modalities, MRI is currently the most widely used and most informative noninvasive technique for evaluating wrist ligaments, although its diagnostic performance remains variable. Artificial intelligence (AI) has emerged as a potential diagnostic adjunct in musculoskeletal imaging. In larger joints, particularly the knee, AI-based models have demonstrated high diagnostic performance and, in some settings, improved reader specificity and overall diagnostic accuracy. In wrist imaging, however, AI applications remain limited and are largely focused on selected tasks. Although early findings are promising, important challenges persist, including small datasets, heterogeneous imaging protocols, limited external validation, and imperfect reference standards. AI is therefore unlikely to improve wrist ligament diagnostics through detection alone. Its real clinical value will likely depend on whether it can help distinguish clinically meaningful pathology from imaging findings of uncertain significance. Level of Evidence IV.