Abstract
Early diagnosis of temporomandibular disorders is challenging. Particularly, intra-articular temporomandibular joint (TMJ) abnormalities can only be confirmed using magnetic resonance imaging (MRI). This study aimed to develop a comprehensive screening method for MRI-detectable TMJ pathologies. We developed an interpretable deep learning framework that leveraged paired open- and closed-mouth TMJ panoramic radiographs and structured clinical metadata. The architecture integrated anatomically guided attention, multimodal clinical features, and ensemble learning for enhanced diagnostic accuracy and interpretability. Across 1355 patients (2710 joints), the best-performing ensemble framework achieved an area under the curve of 0.86, with a balanced classification of MRI-negative and -positive cases. Gradient-weighted Class Activation Mapping visualizations confirmed a consistent focus on the condylar regions, and ablation studies demonstrated the added value of clinical metadata and spatial attention. In conclusion, our prototype workflow can be useful to triage TMJ patients for MRI referral, thus supporting early detection of TMJ abnormalities and timely interventions.
Similar content being viewed by others
Data availability
The datasets generated and/or analyzed during the current study are not publicly available due to the inclusion of patient medical imaging data and related privacy and ethical considerations, but are available from the corresponding author (YP: darkstar@yuhs.ac) on reasonable request.
References
Zieliński, G., Pająk-Zielińska, B. & Ginszt, M. A meta-analysis of the global prevalence of temporomandibular disorders. J. Clin. Med. 13, 1365 (2024).
Alqutaibi, A. Y. et al. Global prevalence of temporomandibular disorders: a systematic review and meta-analysis. J. Oral Facial Pain Headache 39, 48–65 (2025).
Valesan, L. F. et al. Prevalence of temporomandibular joint disorders: a systematic review and meta-analysis. Clin. Oral Investig. 25, 441–453 (2021).
Trize, D. M., Calabria, M. P., Franzolin, S. O. B., Cunha, C. O. & Marta, S. N. Is quality of life affected by temporomandibular disorders?. Einstein 16, eAO4339 (2018).
Schiffman, E. et al. Diagnostic Criteria for Temporomandibular Disorders (DC/TMD) for Clinical and Research Applications: recommendations of the International RDC/TMD Consortium Network* and Orofacial Pain Special Interest Groupdagger. J Oral Facial Pain Headache 28, 6–27 (2014).
Ahmad, M. et al. Research diagnostic criteria for temporomandibular disorders (RDC/TMD): development of image analysis criteria and examiner reliability. Oral Surg. Oral Med. Oral Pathol. Oral Radiol. Endod. 107, 844–860 (2009).
Singer, S. R. & Mupparapu, M. Temporomandibular joint imaging. Dent. Clin. North Am. 67, 227–241 (2023).
Choi, E., Kim, D., Lee, J. Y. & Park, H. K. Artificial intelligence in detecting temporomandibular joint osteoarthritis on orthopantomogram. Sci. Rep. 11, 10246 (2021).
Choi, E. et al. Artificial intelligence-enhanced diagnosis of degenerative joint disease using temporomandibular joint panoramic radiography and joint noise data. Sci. Rep. 15, 1823 (2025).
Kim, D. et al. Expert system for mandibular condyle detection and osteoarthritis classification in panoramic imaging using R-CNN and CNN. Appl. Sci. 10, 7464 (2020).
Almasan, O. et al. Temporomandibular joint osteoarthritis diagnosis employing artificial intelligence: systematic review and meta-analysis. J. Clin. Med. 12, 942 (2023).
Mehta, V., Tripathy, S., Noor, T. & Mathur, A. Artificial intelligence in temporomandibular joint disorders: an umbrella review. Clin. Exp. Dent. Res. 11, e70115 (2025).
Nozawa, M. et al. Automatic segmentation of the temporomandibular joint disc on magnetic resonance images using a deep learning technique. Dentomaxillofac. Radiol. 51, 20210185 (2022).
Manek, M. et al. Temporomandibular joint assessment in MRI images using artificial intelligence tools: where are we now? A systematic review. Dentomaxillofac. Radiol. 54, 1–11 (2025).
Wiese, M. et al. Association between temporomandibular joint symptoms, signs, and clinical diagnosis using the RDC/TMD and radiographic findings in tomograms. J. Orofac. Pain 22, 239–251 (2008).
Manfredini, D., Basso, D., Salmaso, L. & Guarda-Nardini, L. Temporomandibular joint click sound and MRI-depicted disk position: which relationship? J. Dent. 36, 256–260 (2008).
Schiffman, E. & Ohrbach, R. Executive summary of the Diagnostic Criteria for Temporomandibular Disorders. J. Am. Dent. Assoc. 147, 438–445 (2016).
Liu, C. G., Yap, A. U., Fu, K. Y. & Lei, J. The “5Ts” screening tool: enhancements and threshold values for effective TMD identification. Oral Dis 30, 4495–4503 (2024).
Koufos, E. B. et al. The TMD-7 as a brief measure for assessing temporomandibular disorder. Eur. J. Dent. 17, 456–463 (2023).
Gonzalez, Y. M. et al. Development of a brief and effective temporomandibular disorder pain screening questionnaire: reliability and validity. J. Am. Dent. Assoc. 142, 1183–1191 (2011).
Poluha, R. L. et al. Temporomandibular joint disc displacement with reduction: mechanisms and clinical presentation. J. Appl. Oral Sci. 27, e20180433 (2019).
Alexiou, K., Stamatakis, H. & Tsiklakis, K. Evaluation of temporomandibular joint osteoarthritic changes related to age using cone beam CT. Dentomaxillofac. Radiol. 38, 141–147 (2009).
He, K., Zhang, X., Ren, S. & Sun, J. Deep residual learning for image recognition. In Proc. CVPR 770, 778 (2016).
Tan, M. & Le, Q. V. EfficientNet: rethinking model scaling for convolutional neural networks. In Proc. Int. Conf. Mach. Learn. 97, 6105–6114 (2019).
Selvaraju, R. R. et al. Grad-CAM: visual explanations from deep networks via gradient-based localization. In Proceedings of ICCV, 618–626 (IEEE, 2017).
Abd Elaziz, M. et al. Mandibular condyle detection using deep learning and modified mountaineering team-based optimization. Alex. Eng. J. 107, 280–297 (2024).
Bordoni, B. & Varacallo, M. A. Anatomy, head and neck, temporomandibular joint. In StatPearls (StatPearls Publishing, 2023).
Fillingim, R. B. et al. Psychological factors associated with development of TMD: the OPPERA cohort study. J. Pain 14, T75–T90 (2013).
Ulmner, M. et al. Cytokines in temporomandibular joint synovial fluid and tissue in relation to inflammation. J. Oral Rehabil. 49, 599–607 (2022).
Reneker, J., Paz, J., Petrosino, C. & Cook, C. Diagnostic accuracy of clinical tests and signs of temporomandibular disorders. J. Orthop. Sports Phys. Ther. 41, 408–416 (2011).
Ghods, K., Azizi, A., Jafari, A. & Ghods, K. Application of artificial intelligence in clinical dentistry. J. Dent. 24, 356–371 (2023).
Pham, T. Ethical and legal considerations in healthcare AI. R. Soc. Open Sci. 12, 241873 (2025).
Rajmalani, M. & Kumari, S. Role of artificial intelligence in dental diagnostics. JAMDSR 13, 5–12 (2025).
Im, Y. G. et al. Diagnostic accuracy and reliability of panoramic TMJ radiography to detect bony lesions. J. Dent. Sci. 13, 396–404 (2018).
Lee, C., Ha, E.-G., Choi, Y. J., Jeon, K. J. & Han, S.-S. Synthesis of T2-weighted images from proton density images using a GAN in TMJ MRI protocol. Imaging Sci. Dent. 52, 393–398 (2022).
Acknowledgements
This research was supported by the Basic Science Research Program of the National Research Foundation (NRF) of Korea funded by the Ministry of Education (No. RS-2023-00241352) and by the Yonsei University College of Dentistry Fund (No. 6-2023-0062).
Author information
Authors and Affiliations
Contributions
H.-J.J. and Y.P. conceived the study; H.-J.J. and C.L. curated the data; D.J. and C.K. performed the formal analysis; Y.P. acquired funding; H.-J.J. and D.J. conducted the investigation; C.K., S.J.H., and Y.P. developed the methodology; Y.P. managed the project; H.-J.J. and D.J. provided resources; D.J. and C.K. developed the software; Y.P. supervised the study; Y.P. and C.L. validated the data; H.-J.J. and D.J. wrote the original draft; S.J.H., C.L., and Y.P. reviewed and edited the manuscript; all authors read and approved the final manuscript.
Corresponding author
Ethics declarations
Competing interests
The following pending patent application is related to this manuscript: Yonsei University has filed a domestic (Korea) patent application with the application number 10-2025-0112687, covering aspects of the AI-based screening method for temporomandibular joint intra-articular disorders using panoramic radiographic images described in this study. The inventors are Y.P., S.J.H., H.-J.J., C.K., and D.J. The patent application has been filed and is currently pending. The remaining author declares no competing interests.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Jung, HJ., Ju, D., Kim, C. et al. Multimodal deep learning with anatomically constrained attention for screening MRI-detectable TMJ abnormalities from panoramic images. npj Digit. Med. (2026). https://doi.org/10.1038/s41746-026-02378-y
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41746-026-02378-y
