Assessment of human pelvic ring a CT-based statistical shape model of anatomical Principal Variations

To construct a three-dimensional statistical shape model of the pelvis and investigate its three-dimensional morphological changes. We collected CT data from 201 Chinese individuals and used deep learning to reconstruct three-dimensional models of the pelvis automatically. Through three-dimensional model registration, dense correspondence mesh mapping, and the use of statistical shape modeling (SSM) and principal component (PC) analysis methods, we extracted models of variations (MoVs) of pelvic shape changes and statistically compared the shape MoVs between males and females. We analyzed the top 10 principal components of shape variations, which accounted for 86.1% of the total variability. Among them, PC1, PC2, and PC4 showed significant differences between genders (p-values of 0.000, 0.000, and 0.010), accounting for a total variability of 60.1%. PC8 and PC10 demonstrated pelvic asymmetry, accounting for a total variability of 3.8%. We constructed a three-dimensional statistical shape model of the pelvis in Chinese individuals, deepening our understanding of anatomical variations in pelvic morphology. This model can also be further applied in anatomy education and implant design.