A Novel Method for Estimating Three-Dimensional Apical Vertebral Rotation Using Two-Dimensional Coronal Cobb Angle and Thoracic Kyphosis

Publication date: July 2017
Source:Spine Deformity, Volume 5, Issue 4
Author(s): T. Barrett Sullivan, Tracey Bastrom, Fredrick Reighard, Megan Jeffords, Peter O. Newton
Study DesignRetrospective cohort analysis.ObjectivesTo use a large cohort of three-dimensional (3D) spinal reconstructions to create a simple mathematical formula capable of estimating 3D apical vertebral rotation (AVR) based on the correlation with routinely obtained two-dimensional (2D) measurements of scoliosis.Summary of Background DataQuantification of vertebral rotation in AIS using 2-dimensional (2D) imaging is inherently challenging as the axial plane cannot be directly visualized.MethodsA database of 279 3D spinal reconstructions was queried for patients with thoracic major adolescent idiopathic scoliosis (AIS). 2D thoracic Cobb angle, T5–T12 thoracic kyphosis, pelvic incidence, sacral slope, and pelvic tilt were recorded. 3D AVR was calculated for each patient from 3D reconstructions. Patients were divided into development (n = 186) and validation (n = 93) cohorts. Within the development cohort, univariate analysis was performed between 2D measurements and 3D AVR with significance set at p < .05 for inclusion in multivariate analysis. In multivariate analysis, significance was set at p < .01 for inclusion in the final model. Model performance was tested in development and validation cohorts.ResultsOnly 2D thoracic Cobb and T5–T12 thoracic kyphosis had significance in univariate (p < .05) and multivariate analyses (p < .01), meriting inclusion in the final model. 3D AVR (°) = 0.26*(T5–T12 kyphosis) + 0.34*(coronal Cobb) − 5.38. In the development cohort, the model performed well (R = 0.739, r² = 0.54). In testing with the validation cohort, the model proved generalizability (R = 0.703) and had a mean absolute error <5°.ConclusionsThis model is capable of estimating 3D AVR given 2D thoracic Cobb and T5–T12 kyphosis. The accuracy of this method is comparable to previously reported methods of 2D axial rotation measurement. However, this model provides 3D axial rotation and requires no physical instruments, non-standard measurements, or software programs. Such a model is valuable for both routine evaluation of AIS and operative preparation.Level of EvidenceLevel II, diagnostic.



http://ift.tt/2slkQO6