Design: Retrospective study
Introduction: Scoliosis severity is quantified with the Cobb angle on radiographs. Surface topography is a non-invasive assessment of cosmetic features. We proposed new rib cage deformity measurements from full-torso surface topography, but their ability to discriminate among severity subgroups is unknown.
Objective: 1) To determine the discriminating ability of 8 new surface topography features among subjects with different scoliosis severity. 2) Determine the best combination of indices in discriminating among 3 severity subgroups.
Materials and Methods: Torso scans of 87 adolescents (n=29 in each of 3 groups) were categorized into healthy control and mild (Cobb angle <=30) and moderate scoliosis (Cobb angle>30). Seven anatomical landmarks were digitized on each scan and surface features representing rib cage deformity were measured using Matlab.
Results: ANOVAs comparing subgroups suggests that f7 [inferior angle between the lower corner of the scapulas (LCSs) and posterior corner of acromions (PCAs)] and f4 [sagittal angle between the LCS to PCA lines] can discriminate among groups (p<=0.03 for all subgroup comparisons) independantly. Using a Reference Point Logistic regression model, 62% of the subjects were classified correctly using f7 only (23/29 control, 13/29 mild and 18/29 moderate). Adding f4 to the model, only 57% were correctly classified (20, 11 and 18, respectively). Only, 3 subjects were misclassified by two categories.
Conclusion: Only two of the 8 proposed features, f7 and f4, discriminated among groups with different scoliosis severity. F7 was best at discriminating the mild subgroup.
Significance: Two rib cage features may help classify patients into different scoliosis severity subgroups non-invasively.