AI and radiologist assessments on the prevalence of TMJ osteoarthritis using radiographic images: A comparative study

Abstract

Objectives: This study aims to address the significant discomfort and functional impairment associated with temporomandibular joint osteoarthritis (TMJ OA), which negatively impacts the quality of life. It emphasizes the importance of prompt diagnosis and explores the potential of an Artificial Intelligence (AI) system to enhance TMJ OA diagnosis.

Methods: The prevalence of TMJ OA was evaluated using 3 diagnostic tools: the gold standard, the AI model, and an examiner. In total, 132 patients who performed 190 cone-beam computed tomography (CBCT) images were included.

Results: The prevalence of TMJ OA was 62.11% using the gold standard, 63.68% using the AI model, and 58.42% when assessed by the examiner. No gender variation in TMJ OA diagnosis was reported (p-value>0.05). Age variations were reported with the gold standard and the examiner diagnosis. When compared to the gold standard, the AI model had remarkable sensitivity (97.46%) and specificity (91.67%).

Conclusions: The AI model shows promise in enhancing the accuracy of TMJ OA diagnosis, offering potential benefits for early detection and improved patient outcomes.