POPULATION DIFFERENCES IN FEMORAL BONES USING STATISTICAL SHAPE MODELS

Statistical Shape Modeling (SSM) provides a robust framework for analyzing morphological variations in bones across populations, genders, and ethnicities. The Bhattacharyya coefficient, a measure of similarity between probability distributions, has been previously used to compare SSMs constructed from a single dataset with multiple point-set registration variations 1 . However, clinical and biomechanical investigations require evaluating local shape variations between distinct populations. In this study, we developed a vertex-wise Bhattacharyya metric approach to compare SSMs of the full femur bone across three geographically distinct populations viz. India, Belgium, and Switzerland. This method provides a localized assessment of shape differences, the findings of which have the potential to inform the design of population-specific implants, ensuring better fit and functionality in orthopedic applications. We analyzed CT scans of the femur bones from three populations: 80 Belgian, 50 Swiss, and 24 Indian. By generating 3D meshes from all the scans, a non-rigid alignment algorithm was used to establish point-to-point correspondences across all femurs. Using these correspondences, three mean SSMs—one for each population, were constructed. To compare the SSMs, vertex-wise Bhattacharya overlap between the distributions of corresponding points in each model was measured. Five anatomical regions of interest were defined on the mean femur shape, including the femoral head, neck, greater trochanter, shaft, and distal condyles. For each vertex in these regions, the Bhattacharya metric was computed to assess the local similarity between populations. The Bhattacharyya overlap revealed varying degrees of similarity across the three populations, with some regions demonstrating greater morphological divergence than others. Differences in anatomical features contribute to the nuanced variations in shape. The regions with the lowest overlap values will indicate areas most affected by inter-population differences. We will present visualizations of the vertex-wise Bhattacharyya overlaps on the femur, highlighting the anatomical regions with the most significant variations. Additionally, statistical summaries for each region will be reported to quantify the extent of differences between populations. The findings of this study provide valuable insights into morphological variations, with potential applications in orthopedics, biomechanics, and anthropology. By identifying key differences in femoral morphology, this research contributes to the development of personalized and population-specific implants, ensuring better compatibility with anatomical variations. Our methodology holds the promise of being applicable to other anatomical structures to facilitate population-based shape analysis in clinical and research settings.