AI-diagnostic and advanced-manufacturing aspects of temporomandibular joint-related disease

Abstract

The Temporomandibular Joint (TMJ) is the sole bilateral joint in the craniofacial region. Temporomandibular Joint Disease (TMD) typically progresses through three phases: functional disorder, structural disorder, and tissue destruction. As one of the most prevalent orofacial pain disorders, TMD significantly impacting patients' quality of life with symptoms such as chronic pain, restricted jaw movement, and clicking sounds, which can impair speech and mastication.

The disk-condylar structure complexity within TMJ remains a challenging aspect of early diagnosis on magnetic resonance imaging (MRI). Accurate diagnosis of TMD is complicated by its multifactorial nature, variability of symptoms, and overlap with other orofacial conditions. Current MRI-based diagnostic methods often lack consistency and objectivity, which might lead to misdiagnoses and suboptimal treatment outcomes. As such, Artificial Intelligence (AI) has emerged as a promising tool to enhance the accuracy and reliability of TMD(s) diagnosis. Machine learning algorithms such as Convolutional Neural Networks (CNNs) and Fully Connected Network (FCN) can analyze clinic datasets to detect subtle changes in the TMJ structure and function, providing a more precise assessment of disease severity and progression. However, AI-driven TMD diagnostic systems face challenges in sensitivity and specificity.

Further, in advanced TMD stage, conservative treatments may become ineffective, necessitating surgical interventions such as arthroscopy, arthroplasty, or total joint replacement. These procedures are often complex and require patient-specific planning and execution to achieve optimal results. In particular, AI-driven computer-aided design (CAD) and computer-aided manufacturing (CAM) systems enable the creation of customized prosthetic devices tailored to the anatomical and functional needs of TMD patients. Despite these advances, gaps remain in integrating AI technologies into clinical practice, particularly in transitioning from diagnostics to treatment. Current research often isolates diagnostic or therapeutic aspects without a comprehensive approach linking the two stages. This study aims to bridge this gap by developing an integrated AI-based diagnostic and therapeutic framework for TMD. The objectives include evaluating AI algorithms' effectiveness in diagnosing TMD stages, and assessing the feasibility of TMJ prosthesis design.

This study developed an AI-based MRI analysis system, achieving a sensitivity of 98.8%, a specificity of 98.8%, an accuracy of 98.6%, and an F1-score of 98.7% in detecting abnormalities related to TMD. In addition, particularly for advanced treatment, cobalt-chromium (Co-Cr) alloy is commonly used for TMJ prosthesis. However, its biocompatibility and toxicity have not been comprehensively evaluated. This study examines various cobalt-chromium alloys both in vitro and in vivo aspects, comparing biocompatibility and toxicity across different manufacturing and polishing methods. Findings indicate that 3D-printed Co-Cr alloys have higher surface roughness but lower ion release and cytotoxicity, while cast alloys show greater cytotoxicity but lower genotoxicity and no significant skin sensitization.

In conclusion, the application of AI-driven diagnostic system together with the 3D printed Co-Cr with specific surface treatments not only improves early detection but also aids in guiding clinicians towards personalized and effective management strategies for TMD.