Repeatability and agreement of planar lumbar range of motion measured by dynamic fluoroscopy and optical motion capture in an ovine cadaveric model

Introduction Lumbar range of motion is an important biomechanical parameter for understanding spinal function. However, the repeatability, measurement error, and between-method agreement of dynamic fluoroscopy and optical motion capture remain insufficiently characterized in controlled cadaveric spine testing. Methods Planar lumbar kinematics were measured using dynamic fluoroscopy within a dynamic X-ray system (DXRS) and optical motion capture with bone-fixed markers (OMCS) in six fresh-frozen ovine lumbar spine specimens, each comprising a continuous L1-L5 segment, during quasi-static flexion-extension and lateral bending. Three repeated trials were performed for each condition. Repeatability was assessed using the intraclass correlation coefficient (ICC(2,1)), the standard error of measurement (SEM), and the minimum detectable change at the 95% confidence level (MDC95). Between-method agreement was evaluated using a repeated-measures Bland-Altman approach with specimen-level clustering, exploratory proportional-bias assessment, and specimen-level bootstrap confidence intervals (CIs) (10,000 replicates) for the bias and limits of agreement (LoA). Results ICC point estimates were generally high within each method across motion conditions and vertebral levels; however, given the modest specimen sample (n = 6), the corresponding 95% CIs were wide and these estimates are reported as preliminary, condition-specific values. OMCS showed numerically smaller SEM and MDC95 than DXRS in this dataset, but specimen-level bootstrap intervals indicated uncertainty around most between-method differences, so these descriptive differences should not be interpreted as evidence of general method superiority. Between-method analysis showed a small positive OMCS-minus-DXRS bias with condition-dependent LoA, and these agreement estimates should likewise be interpreted as preliminary and condition-specific. Conclusion In this controlled ovine cadaveric model, both DXRS and bone-fixed OMCS provided repeatable measurements of projected planar lumbar motion. OMCS showed numerically smaller measurement-error indices in this dataset and under the present bone-fixed ex vivo ovine configuration; these descriptive, configuration-specific differences do not imply general method superiority, clinical equivalence, or interchangeability with routine human skin-marker motion capture. Cross-device comparisons should account for systematic bias, LoA, and uncertainty related to repeated-measures clustering. The reported error and agreement estimates may help inform method selection and interpretation of longitudinal change in preclinical spine biomechanics studies.