The mechanical behaviour of the craniomaxillofacial skeleton (CMFS) is poorly understood due to its intricate geometry, thin bone structure, high sensitivity to loads and boundary conditions, and complex muscle loading. Yet, this limited knowledge is currently the foundation for CMFS reconstruction. Towards reducing post-surgical complications, an understanding of load transmission patterns and mechanical stability is needed to guide optimum placement and identify minimum required hardware for CMFS fracture fixation. Similarly, CMF soft tissue reconstruction can be challenging after severe traumatic injuries due to the limited access to the 3D face shapes pre-injury. Artificial intelligence (AI) based methods have been used to create facial anatomies and may be useful in understanding the relationship between soft tissue and the underlying bone structure that could aid in CMFS reconstruction. This thesis seeks to address CMF reconstruction by developing detailed intact and post-fracture fixation CMFS finite element (FE) models and an AI skull-face shape model that allows relative estimation of soft tissue geometry using the underlying bone geometry.

Intact and post-fracture fixation FE models were developed by integrating information from multimodality imaging to accurately model bone, muscle, and muscle fiber architecture. Results show that muscle modeling using simplified link elements yields more diffuse patterns of high-strain than achieved with 3D muscle modeling. Post-fracture fixation FE results suggest that two-point fixation provides sufficient zygomaticomaxillary complex (ZMC) fracture stability and results in a strain pattern most similar to the intact bone. This aligns with the growing body of evidence that three-point fixation may not be needed to adequately stabilize ZMC fractures. Estimating facial shape from the underlying skull geometry is impacted by multiple factors (including age, sex, BMI, and race). Using large publicly available head CT databases, an AI model was developed to predict an individual’s face shape using the underlying bone shape and additional metadata (validation set DC=0.92). Overall, through a combination of imaging, computational modeling and experimental methods, this thesis has enabled quantification of load transmission patterns through the CMFS and accurate estimation of pre-traumatic CMF geometry, providing insights relevant to clinical reconstruction.