Accuracy of the Artificial Intelligence to Locate the Temporomandibular Disc in MRI

Temporomandibular disorder (TMD) is a term used to describe the morphological and/or functional changes that affect the masticatory system. Internal derangements of the TMJ are among the most prevalent articular subtypes of TMD. TMD diagnoses are based on clinical and imaging examinations, mostly magnetic resonance imaging (MRI). Artificial intelligence (AI) has emerged as a method for optimizing determination of the position of the articular disc of the TMJ using MRI. This proof-of-concept study aimed to explore the reliability and accuracy of applying AI to optimize TMD diagnoses. This retrospective cross-sectional study comprised 459 images from 67 participants.. Neural networks and deep learning were developed. The study showed that the coefficient of agreement () between AI-analyzed MRI and TMJ MRI evaluated by specialist was 0.63. AI provided high reliability (84,6%) to locate the articular disc in MRI. Deep learning program revealed a sensitivity of 86.2% and specificity of 76.9% on the 1,000-step test. The 50,000-step test showed a sensitivity of 87.7% and specificity of 76.9%. ROC curve (AUC > 0.8) emphasized the high diagnostic accuracy. These findings demonstrated that AI is effective in optimizing the location of the TMJ disc through MRI.