Comparison of the performance of a Three-Dimensional Body Scanner and radiography in evaluating adult scoliosis

Background. Adult scoliosis, which is characterized by a persistent lateral deviation of the spine of at least 10◦ in the frontal plane along with vertebral rotation in adulthood, can result from various causes, including degenerative changes, untreated childhood scoliosis, spinal trauma, and prior surgeries. Traditionally, spinal curvature is assessed by measuring the Cobb angle via radiographic imaging; however, concerns over radiation exposure have prompted exploration of alternative diagnostic tools. This study aims to evaluate the effectiveness of a newly developed three-dimensional (3D) body scanner, equipped with 12 depth cameras, in assessing spinal alignment and measuring the Cobb angle in patients with adult scoliosis, in comparison with radiographic imaging. Methods. In this prospective cohort study, 40 patients with adult scoliosis—both idiopathic and degenerative—underwent evaluation using both radiographic imaging and 3D body scanning. Cobb angles were measured by both methods. Pearson and Spearman’s rank correlation coefficients were calculated to assess the linear and monotonic relationships between measurements. Measurement accuracy was quantified using the mean bias from Bland–Altman analysis and spatial agreement of spinal positions was further evaluated using the Intersection over Union (IoU) metric. Univariable and multivariable regression analyses were performed to assess whether body habitus (body mass index, waist circumference, waist-to-height ratio, age, and sex) influenced the absolute error between 3D body scanner-predicted and radiographic Cobb angles. Results. Cobb angle measurements obtained from 3D body scanning were highly correlated with those from radiography (Pearson r = 0.92, P < 0.001; Spearman ρ = 0.85, P < 0.001), indicating strong linear and monotonic agreement. Bland–Altman analysis showed a small mean bias of −1.06 (95% limits of agreement: −10.25 to 8.12). The average IoU was 0.89, indicating substantial spatial agreement in spinal position predictions. Importantly, obesity indices (body mass index, waist circumference, waist-to-height ratio) were not significantly associated with the absolute error between 3D body scanner-predicted and radiographic Cobb angles in either univariable or multivariable analyses. Conclusions. The 3D body scanner exhibits promise for assessing spinal alignment and measuring the Cobb angle in patients with adult scoliosis, offering a reliable alternative to traditional radiographic methods. Its accuracy was not affected by obesity-related indices, supporting its applicability across diverse patient body types. Future research should focus on refining scanning protocols and integrating patient-reported outcomes to enhance clinical utility. Copyright 2026 Kuo et al.