ARTIFICIAL INTELLIGENCE FOR DIAGNOSTIC ASSESSMENT OF SAGITTAL SPINAL ALIGNMENT: A SCOPING REVIEW

INTRODUCTION Sagittal spinal alignment assessment is critical for diagnosing spinal deformities, planning surgical interventions, and monitoring treatment outcomes. Conventional measurement techniques require considerable expertise and are susceptible to measurement error. Artificial intelligence (AI) technologies offer the potential to standardise diagnostic measurements, reduce variability, and support clinical decision-making in sagittal spine evaluation. AIM This scoping review examined the application of AI in the diagnostic assessment of sagittal spinal alignment, evaluating the scope of current evidence, the measurement parameters addressed, and diagnostic performance. METHOD A systematic search was performed across PubMed, Embase, Scopus, and Web of Science databases up to October 2025. Studies were eligible if they described AI-based systems designed to support diagnostic assessment of sagittal spinal parameters, including automated measurement tools and diagnostic classification systems. Extracted data included the AI methodology, the spinopelvic parameters measured, the validation approaches, and the reported accuracy compared with expert measurements. RESULTS Twenty-five studies were included in this review. AI models predominantly utilised deep learning architectures, including convolutional neural networks and encoder-decoder frameworks. Key diagnostic parameters assessed included pelvic incidence, sacral slope, pelvic tilt, lumbar lordosis, thoracic kyphosis, and sagittal vertical axis. Several studies validated AI measurements against expert clinicians, demonstrating comparable or superior reliability. Applications spanned cervical, thoracic, and lumbar regions, with some platforms achieving fully automated analysis pipelines. CONCLUSION AI demonstrates considerable promise for enhancing diagnostic accuracy and consistency in sagittal spinal alignment assessment. Future research should prioritise multi-centre validation, integration into clinical workflows, and comparative effectiveness studies to establish AI as a reliable diagnostic adjunct.