Reliability and Reproducibility of Facet Joint Measurements in the Thoracolumbar Spine

The objective is to evaluate the reliability and reproducibility of CT-based assessment methods: * Reliability and reproducibility of grading FJ degenerative changes using the Pathria scale; * Reliability and reproducibility of grading IvD degenerative changes, including tendency toward ankylosis formation; * Reliability and reproducibility of quantitative measurements of FJ space width (mean, minimum, and maximum values); * Reliability and reproducibility of FJ angulation measurements in the axial plane, including assessment of FJ tropism. At least six raters will participate, including three neurosurgeons and three CT specialists. CT scans will be selected in random order from eligible imaging studies. Included scans must cover vertebral levels T9-L3, be obtained in patients aged 18 to 65 years, and be free from imaging artifacts at the levels of interest. Exclusion criteria for scans include vertebral bone density below 100 HU, systemic bone or joint disorders associated with changes in FJs or IvDs, previous spinal surgery, spinal fractures involving injury to posterior spinal structures, and participation in other clinical studies related to spinal disorders or spinal trauma. Sample size for the quantitative reproducibility analysis was calculated in R using the ICC.Sample.Size package. Assuming a minimum acceptable intraclass correlation coefficient (ICC) of 0.60, an expected ICC of 0.75, alpha 0.05, power 80%, and 6 raters, the minimum required sample size was 47 FJs; after allowing for 10% data loss, the target sample size was 53 FJs. Because up to 12 FJs can be evaluated per patient, this corresponds to at least 5 patients. Sample size for reproducibility of the 4-grade Pathria classification was calculated in R using the kappaSize package. Assuming a minimum acceptable kappa of 0.50, an expected kappa of 0.60, alpha 0.05, power 80%, and 6 raters, the minimum required sample size was 118 observations; after allowing for 10% loss of evaluable data, the target sample size was 130 FJs, corresponding to at least 11 patients. Intra- and inter-rater agreement will be assessed using Fleiss' kappa for ordinal variables and the intraclass correlation coefficient (ICC) for continuous variables.