Two-stage deep learning approach for screening for anterior disk displacement of the temporomandibular joint using orthopantomograms.

This study aimed to develop a deep learning-based approach for the initial screening of anterior disk displacement (DD) of the temporomandibular joint (TMJ) using orthopantomograms (OPG). A two-stage deep learning model was proposed: first, regions of interest were detected on orthopantomogram images using YOLOv5s; second, a classification model based on ResNet-18 and DANet was trained on magnetic resonance imaging–verified labels to identify TMJ DD. Diagnostic performance was evaluated through multi-class and binary classification analyses. In three-class classification (normal disk position, DD with reduction, and DD without reduction), the model achieved an overall accuracy of 69.22%. It performed well in identifying normal disk position (F1 score: 75.53%) and DD without reduction (F1 score: 77.49%) but showed lower performance in detecting DD with reduction (F1 score: 47.33%). In the binary classification, where DD with reduction and DD without reduction were combined into a single class, the model demonstrated improved accuracy (80.78%), sensitivity (81.93%), and specificity (78.89%). Beyond classification, the model also exhibited potential in estimating DD severity (R2 = 0.4768). Given its strong ability to differentiate between normal disk position and DD in a binary classification setting, this tool has the potential to serve as an initial screening tool for TMJ DD in clinical practice.