Imaging pathways in spondyloarthritis: integrating radiography, ultrasonography, magnetic resonance imaging, low-dose computed tomography, and artificial intelligence methods

Abstract Spondyloarthritis (SpA) refers to a family of chronic inflammatory rheumatic conditions characterized by axial and/or peripheral manifestations. Early detection of SpA is crucial for improving long-term patient outcomes and necessitates a refined diagnostic algorithm. This literature review addresses current recommendations for imaging approaches in SpA, proposes a contemporary diagnostic algorithm for suspected axial SpA, and discusses current and emerging applications of artificial intelligence (AI) in diagnosis and management. A comprehensive literature search of PubMed, Embase and Scopus was performed for studies published between January 2010 and August 2025. Relevant English-language studies on imaging modalities and AI applications in SpA were included after independent screening. The implementation of advanced imaging techniques—such as low-dose computed tomography (CT) for detailed structural assessment and standardized magnetic resonance imaging (MRI) protocols for detecting inflammatory changes—has improved the diagnostic evaluation of sacroiliac joints. Incorporating clinical features and modality-specific strengths helps tailor imaging choices to individual patients with suspected SpA. In clinical practice, MRI may be considered for early detection of sacroiliitis—especially in younger patients and those with short symptom duration—whereas conventional radiography continues to serve as the recommended first-line imaging modality in many diagnostic pathways. Low-dose CT should be reserved for selected cases, such as inconclusive MRI findings, contraindications to MRI, limited MRI availability, or a specific need to assess structural damage. Advances in AI, particularly in deep learning, have had a remarkable impact on medical research. Despite existing limitations, such as costs of deployment and medico-legal considerations, their role in rheumatological imaging is being actively investigated. Deep learning-based models trained on radiographic, CT and MRI datasets have demonstrated progressively greater precision in detecting sacroiliitis, becoming a powerful tool that complements human judgement. Prospective strategies integrating multimodal imaging, AI-assisted interpretation, and prognostic assessment may enhance diagnostic accuracy and provide personalized therapeutic solutions in SpA.