124. Dense Phenotyping Reveals Sexually Dimorphic Craniofacial Skeletal Morphology With High Fidelity

PURPOSE: Accurate understanding of craniofacial bone morphology (CBM) is crucial for facial gender surgery (FGS), where subtle facial structure differences can significantly affect patient self-perception and dysphoria. Current methods rely on manual (sparse) landmarking with a few dozen to hundreds of semi-landmarks. This approach is insufficient to capture complex CBM and can be time-consuming and imprecise. More robust landmarking techniques with higher density and more automation are therefore needed to fully define CBM and enhance surgical planning. This study applies a densely annotated craniofacial template mask and registration pipeline, both previously validated for high anatomical fidelity, to analyze sexual dimorphism, aiming to improve FGS precision and enable the projection of potential surgical outcomes for individualized visualization. METHODS: 3D skull meshes from CT images of 15 males and 15 females were extracted. A mask with 9,999 symmetrized vertices was applied to quantify CBM. Application involved Procrustes alignment with manual landmarks, iterative closest point refinement, and non-rigid registration via viscoelastic deformations. After excluding 3,291 non-skull vertices (e.g., eye sockets), we aligned skull vertices to generate average skull shapes per sex. Vertex-wise differences were computed using independent t-tests, correcting for multiple comparisons. Principal Component Analysis (PCA) identified regions contributing most to craniofacial variance. Heatmaps projected onto 3D skull renderings visualized significant sex-shape differences. RESULTS: Our analysis revealed statistically significant CBM differences between male and female skulls across all 6,708 anatomical vertices. Heatmaps overlaid on average skulls visualized sexual dimorphism magnitude and locations, highlighting areas for surgical modification. Significant variations were found in the glabella, nasal bone, maxilla, and chin projection. PCA showed separation between male and female skulls along the first two principal components, accounting for 33.15% of morphological variance. CONCLUSION: We achieved greater precision in quantifying craniofacial sexual dimorphism, identifying specific morphological differences critical for surgical planning in FGS. Previous studies, which relied on sparse manual landmarking with only a dozen to a few hundred points, often identified general areas of dimorphism around the landmarks. However, they lacked the resolution to capture subtleties in the landmarks’ surroundings. Our analyses revealed more precise CBM differences, particularly around those generalized areas. These findings provide quantitative data that can be applied to individual patients’ 3D models to develop preoperative planning tools, enabling surgeons to plan patient-specific approaches targeting regions with the greatest sexual dimorphism and allowing for specific bony volume reduction. Future applications include developing interactive visualization platforms where patients can see various craniofacial skeletal outcome scenarios, aiding in making informed decisions. Expanding datasets to include more diverse populations will also enable more inclusive and accurate predictions. Refinement of virtual surgical planning tools based on these findings can improve patient understanding and participation in shared decision-making, potentially improving outcomes and patient satisfaction.