A Comprehensive Review on Detection of TMJ Osteoarthritis

Abstract

Temporomandibular joint osteoarthritis (TMJOA) is a degenerative disorder that causes pain, functional impairment, and reduced quality of life. Traditional diagnostic methods, such as manual radiographic assessments, are time consuming and prone to subjectivity. Recent advancements in artificial intelligence (AI), particularly deep learning models, offer promising solutions for automated and accurate TMJOA detection. This review comprehensively examines AI-based approaches, including Convolutional Neural Networks (CNNs), YOLO variants, and hybrid architectures, for diagnosing TMJOA using imaging modalities like panoramic radiographs, cone-beam computed tomography (CBCT), and magnetic resonance imaging (MRI). Key studies demonstrate that AI models achieve high accuracy in condylar segmentation, bone deformation classification, and early OA detection, often outperforming traditional methods. However, challenges such as dataset limitations, model generalizability, and clinical validation remain. The integration of AI with biomechanical and clinical data further enhances diagnostic precision, supporting personalized treatment strategies. This review highlights AI’s transformative potential in TMJOA diagnosis while emphasizing the need for standardized protocols and multicenter collaborations to improve clinical adoption.