Structural hip disease exists along a continuum; however, this spectrum is difficult to capture using traditional two-dimensional measurements alone. Demographic variables and other clinical factors may influence morphology, diagnosis and treatment decisions, but are often not factored into traditional methods for evaluating hip morphology. Statistical shape modeling (SSM), a powerful tool for objective morphometric analysis, could address these challenges by capturing the morphological variation of the full bone and enabling direct comparison of shape across a cohort of individuals. To demonstrate the utility of SSM in evaluating patients with structural hip disease, we utilized SSM to assess differences in overall pelvis morphology in individuals from different diagnosis and demographic groups. The cohort included 73 female controls without a history of hip pain and 75 female DDH patients (48 Japanese, 27 Non-Japanese). Hemi-pelvis segmentations were reconstructed from volumetric images then imported into ShapeWorks 6.5.1. 4,096 correspondence particles were automatically placed on the surfaces using entropy-based optimization and a Procrustes analysis was used to remove the effect of scale on particle position. A combination of dimensionality reduction techniques and descriptive statistics were used to determine where the morphology varied the most between controls and DDH patients as well as between DDH patients from different demographic groups. Interestingly in both comparisons, the location of the most significant variation in pelvis morphology did not occur in the acetabulum, but instead occurred in regions of muscle attachment points. In the comparison between controls and DDH patients we were still able to identify differences in the lateral extent of the acetabulum, confirming we were able to capture the same differences observed from classic radiographic measurements. The patient groups had the most variation in the anterior and posterior extent of the acetabulum. These findings highlight SSMs capabilities to describe morphological differences that extend beyond conventional radiographic assessment.
Braun et al. (Thu,) studied this question.