tsta

Home / سی و دومین کنفرانس ملی و دهمین کنفرانس بین المللی مهندسی زیست پزشکی ایران

An AI-Assisted Approach to Patient-specific 3D Modeling and Stress Analysis of the Temporomandibular Joint from CBCT Images

Authors :

MOHAMMAD Akhlaghi¹ Masoud Shariat panahi² Sina Salehpour³ Morad Karimpour⁴ Hadi Ghatan Kashani⁵

1- دانشگاه تهران 2- دانشگاه تهران 3- university of maryland baltimore 4- دانشگاه تهران 5- دانشگاه تهران

Keywords :

Temporomandibular joint،finite element analysis،CBCT،patient-specific modeling،deep learning segmentation

Abstract :

Accurate assessment of temporomandibular joint (TMJ) biomechanics requires patient-specific models that capture anatomical variability with high fidelity. This study presents an AI-assisted method for biomechanical modeling and stress analysis of the TMJ, integrating deep learning–based segmentation and 3D reconstruction from cone-beam computed tomography (CBCT). CBCT scans are processed using a pretrained UNET-R network, fine-tuned with a small dataset to segment the mandible, maxilla, and temporal bone. The segmented geometries are reconstructed into 3D models, converted into volumetric finite element (FE) meshes, and combined with a modeled TMJ disc representing the soft tissue between the mandibular condyle and maxilla. The FE model is then subjected to simulated incisal and unilateral molar loading to estimate stress distribution within the disc. The proposed method enables rapid and reproducible generation of patient-specific TMJ biomechanical models, reducing manual segmentation effort and computational cost while preserving clinical accuracy. This approach demonstrates strong potential for personalized diagnosis, treatment planning, and biomechanical research on TMJ disorders.