Automated Diagnosis of TMJ Anterior Disc Displacement Using Deep Learning
Author Information
Author(s): Yu Yue, Wu Shu Jun, Zhu Yao Min
Primary Institution: Shenzhen Stomatology Hospital, Affiliated to Shenzhen University
Hypothesis
This study aimed to develop a deep learning-based method for interpreting MRI scans of temporomandibular joint anterior disc displacement and to formulate an automated diagnostic system for clinical practice.
Conclusion
The deep learning models exhibited high accuracy in detecting TMJ anterior disc displacement and provided interpretable, visualized predictive results.
Supporting Evidence
- Classification models demonstrated high precision rates above 92%.
- The segmentation-based model outperformed the non-segmentation model in overall and category-specific metrics.
- Models achieved a mean Average Precision (mAP) of 94.86% in internal validation tests.
Takeaway
Researchers created a computer program that can look at MRI pictures of jaw joints and tell if there's a problem with the disc inside, helping doctors make better diagnoses.
Methodology
The study used deep learning models to identify regions of interest, segment TMJ structures, and classify TMJ anterior disc displacement using MRI datasets from two hospitals.
Potential Biases
The model may not fully capture the diversity of TMJ conditions across different populations or healthcare settings.
Limitations
The dataset was limited to two centers, which may not reflect variations in imaging techniques across different medical institutions.
Participant Demographics
Patients aged 19–40 years (mean age 28.9 ± 6.4 years) with TMJ-related symptoms.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website