| description abstract | Abstract. This study presents a patient-specific parametric model of the temporomandibular joint, designed to be semi-automated, and reproducible for multiple patients. The numerical model is used to evaluate mandibular stress distribution under different interaction and loading conditions. The main contribution is to demonstrate the feasibility of an integrated, fast, and streamlined workflow for generating accurate biomechanical models tailored to each patient. The proposed method, which relies on standard clinical images complemented by artificial intelligence-assisted segmentation of bone and muscles, enables the integration of patient-specific anatomical features and mechanical variability. A finite element model of the skull, mandible, teeth, and articular disks was constructed from calibrated computed tomography data. Material properties were automatically assigned using Hounsfield units, distinguishing between cortical bone, cancellous bone, and dental tissue. Sensitivity of key modeling parameters (mesh density, material, friction coefficients, muscle force vectors) was evaluated using abaqus/standard. Hounsfield-units-driven material assignment provides a Young modulus distribution aligned with the literature, while maintaining patient specificity. Artificial intelligence-based muscle reconstruction reveals that stress fields stabilize with increased directional vector refinement, reinforcing biomechanical accuracy and confirming the necessity of multivector muscle loading. This patient-specific parametric model accurately reproduces the distribution of mandibular stresses and offers a promising tool for surgical planning, pathology simulation, and the evaluation of personalized treatment strategies. | |
