Efficacy of a Deep Learning System for Automatic Analysis of the Mandibular Canal Type on Panoramic Radiographs

Abstract

Accurate anatomical variant detection is critical in clinical diagnostics, yet disparities in imaging modalities often challenge reliable assessment. In dentistry, panoramic radiographs (PRs) are widely used for mandibular canal evaluation, but their reported detection rates for bifid variants (0.038%–1.98%) fall far below those of cone-beam computed tomography (CBCT; 10%–66%), highlighting a need for improved diagnostic tools. Here, we address this gap by developing a deep learning-based tri-comparison expertise decision (TED) system to automate mandibular canal variant classification on PRs. Using retrospective data from 442 mandible sides (279 participants, aged 18–32 years), we validated PRs against CBCT ground truth and decomposed multi-class classification into pairwise comparisons with an “Another” class to enhance discrimination of anatomically similar variants. Here we show that the TED system achieved superior diagnostic accuracy (0.701, 95% CI: 0.674–0.728) and AUROC (0.854, 95% CI: 0.824–0.884) compared to assessments by five experienced dentists (highest accuracy: 0.683; AUROC: 0.810), while also revealing strikingly low inter-rater agreement among experts (Fleiss' kappa = 0.046). These results demonstrate that the TED approach not only outperforms manual evaluations but also provides consistent, cost-effective automation of a task prone to human variability. By bridging the performance gap between PRs and CBCT, this tool offers a practical solution for preoperative risk assessment in dental practice. Broader validation across diverse clinical settings could further solidify its role in improving diagnostic workflows and patient outcomes.

Keywords

cone beam computed tomography / deep learning / inferior alveolar nerve / mandibular canal variants / panoramic radiograph / tri-comparison expertise decision

References

[1] M. S. Hur, H. C. Kim, S. Y. Won, et al., “Topography and Spatial Fascicular Arrangement of the Human Inferior Alveolar Nerve,” Clinical Implant Dentistry and Related Research 15, no. 1 (2013): 88–95.

[2] P. Worthington, “Injury to the Inferior Alveolar Nerve During Implant Placement: A Formula for Protection of the Patient and Clinician,” International Journal of Oral & Maxillofacial Implants 19, no. 5 (2004): 731–734.

[3] P. R. Genú and B. C. E. Vasconcelos, “Influence of the Tooth Section Technique in Alveolar Nerve Damage After Surgery of Impacted Lower Third Molars,” International Journal of Oral and Maxillofacial Surgery 37, no. 10 (2008): 923–928.

[4] P. Coulthard, E. Kushnerev, J. M. Yates, et al., “Interventions for Iatrogenic Inferior Alveolar and Lingual Nerve Injury,” Cochrane Database of Systematic Reviews 2014, no. 4 (2014): CD005293.

[5] M. Naitoh, Y. Hiraiwa, H. Aimiya, et al., “Bifid Mandibular Canal in Japanese,” Implant Dentistry 16, no. 1 (2007): 24–32.

[6] B. Friedland, B. Donoff, and T. B. Dodson, “The Use of 3-Dimensional Reconstructions to Evaluate the Anatomic Relationship of the Mandibular Canal and Impacted Mandibular Third Molars,” Journal of Oral and Maxillofacial Surgery 66, no. 8 (2008): 1678–1685.

[7] E. Olivier, “The Inferior Dental Canal and Its Nerve in the Adult,” British Dental Journal 49, no. 5 (1928): 356–358.

[8] K. T. Wolf, E. J. Brokaw, A. Bell, and A. Joy, “Variant Inferior Alveolar Nerves and Implications for Local Anesthesia,” Anesthesia Progress 63, no. 2 (2016): 84–90.

[9] S. Maekawa, M. Nagata, Y. Matsushita, R. S. Tubbs, and J. Iwanaga, “An Unusual Anatomical Variation of the Inferior Alveolar Nerve,” Anatomy & Cell Biology 53, no. 4 (2020): 519–521.

[10] A. Kuczynski, W. Kucharski, A. Franco, F. H. Westphalen, A. A. S. de Lima, and Â. Fernandes, “Prevalence of Bifid Mandibular Canals In Panoramic Radiographs: A Maxillofacial Surgical Scope,” Surgical and Radiologic Anatomy 36, no. 9 (2014): 847–850.

[11] M. H. Kalantar Motamedi, F. Navi, and N. Sarabi, “Bifid Mandibular Canals: Prevalence and Implications,” Journal of Oral and Maxillofacial Surgery 73, no. 3 (2015): 387–390.

[12] S. Kasabah and Y. Modellel, “Classification of Bifid Mandibular Canals in the Syrian Population Using Panoramic Radiographs,” Eastern Mediterranean health journal = La revue de sante de la Mediterranee orientale = al-Majallah al-sihhiyah li-sharq al-mutawassit 19 Suppl 3, no. Suppl 3 (2014): 178–183.

[13] D. Göller Bulut, G. Kartal Yalçın, Z. Tanrıseven, B. Taşkın, and B. Aydın, “Prevalence and Topography of Bifid and Trifid Mandibular Canal in Turkish Western Anatolia Population: Evaluation of the Inferior Alveolar Canal With Cbct,” Surgical and Radiologic Anatomy 46, no. 10 (2024): 1663–1672.

[14] A. Cuozzo, I. S. Vincenzo, M. Boariu, et al., “Prevalence and Anatomical Characteristics of Bifid and Trifid Mandibular Canals: A Computer Tomography Analysis,” Oral health & preventive dentistry 22 (2024): 301–308.

[15] S. Samieirad, M. Aryana, A. Mazandarani, et al., “Prevalence of Bifid Mandibular Canal: A Systematic Review and Meta-Analysis,” World Journal of Plastic Surgery 12, no. 2 (2023): 11–19.

[16] Y. Alali, W. A. Mohammed, M. Alabulkarim, A. Alshahrani, and A. Almawh, “Assessment of Bifid Mandibular Canals Using Cone Beam Computed Tomography in General Population: A Retrospective Evaluation,” European Review for Medical and Pharmacological Sciences 28, no. 5 (2024): 1741–1750.

[17] N. M. Aung and K. K. Myint, “Bifid Mandibular Canal: A Proportional Meta-Analysis of Computed Tomography Studies,” International Journal of Dentistry 2023 (2023): 1–23.

[18] A. Auluck and K. M. Pai, “Trifid Mandibular Nerve Canal,” Dentomaxillofacial Radiology 34, no. 4 (2005): 259.

[19] J. Iwanaga, Y. Takeshita, Y. Matsushita, M. S. Hur, S. Ibaragi, and R. S. Tubbs, “What Are the Retromolar and Bifid/Trifid Mandibular Canals as Seen on Cone-Beam Computed Tomography? Revisiting Classic Gross Anatomy of the Inferior Alveolar Nerve and Correcting Terminology,” Surgical and Radiologic Anatomy 44, no. 1 (2022): 147–156.

[20] F. Schwendicke, W. Samek, and J. Krois, “Artificial Intelligence in Dentistry: Chances and Challenges,” Journal of Dental Research 99, no. 7 (2020): 769–774.

[21] S. Corbella, S. Srinivas, and F. Cabitza, “Applications of Deep Learning in Dentistry,” Oral Surgery, Oral Medicine, Oral Pathology and Oral Radiology 132, no. 2 (2021): 225–238.

[22] F. Carrillo-Perez, O. E. Pecho, J. C. Morales, et al., “Applications of Artificial Intelligence in Dentistry: A Comprehensive Review,” Journal of Esthetic and Restorative Dentistry 34, no. 1 (2022): 259–280.

[23] M. Alahmari, M. Alahmari, A. Almuaddi, et al., “Accuracy of Artificial Intelligence-Based Segmentation in Maxillofacial Structures: A Systematic Review,” BMC Oral Health 25, no. 1 (2025): 350.

[24] W. Semper-Hogg, A. Rau, M. A. Fuessinger, et al., “Deep Learning-Based Segmentation of the Mandibular Canals in Cone-Beam CT Reaches Human-Level Performance,” Dentomaxillofacial Radiology 54, no. 4 (2025): 279–285.

[25] G. Dot, A. Chaurasia, G. Dubois, et al., “Dentalsegmentator: Robust Open Source Deep Learning-Based CT and CBCT Image Segmentation,” Journal of Dentistry 147 (2024): 105130.

[26] E. T. Yasin, M. Erturk, M. Tassoker, and M. Koklu, “Automatic Mandibular Third Molar and Mandibular Canal Relationship Determination Based on Deep Learning Models for Preoperative Risk Reduction,” Clinical Oral Investigations 29, no. 4 (2025): 203.

[27] N. A. Barnes, W. Dkhar, S. S, Y. Chhaparwal, V. Mayya, and H. Rc., “Automated Classification of Mandibular Canal in Relation to Third Molar Using CBCT Images,” F1000Research 13 (2024): 995.

[28] Y. Jiang, H. T. Sun, Z. Luo, J. Wang, and R. P. Xiao, “Efficacy of a Deep Learning System for Automatic Analysis of the Comprehensive Spatial Relationship Between the Mandibular Third Molar and Inferior Alveolar Canal on Panoramic Radiographs,” Oral Surgery, Oral Medicine, Oral Pathology and Oral Radiology 139, no. 5 (2025): 612–622.

[29] S. Sukegawa, F. Tanaka, T. Hara, et al., “Deep Learning Model for Analyzing the Relationship Between Mandibular Third Molar and Inferior Alveolar Nerve in Panoramic Radiography,” Scientific Reports 12, no. 1 (2022): 16925.

[30] T. Zhu, D. Chen, F. Wu, F. Zhu, and H. Zhu, “Artificial Intelligence Model to Detect Real Contact Relationship Between Mandibular Third Molars and Inferior Alveolar Nerve Based on Panoramic Radiographs,” Diagnostics 11, no. 9 (2021): 1664.

[31] K. Takebe, T. Imai, S. Kubota, A. Nishimoto, S. Amekawa, and N. Uzawa, “Deep Learning Model for the Automated Evaluation of Contact Between the Lower Third Molar and Inferior Alveolar Nerve on Panoramic Radiography,” Journal of Dental Sciences 18, no. 3 (2023): 991–996.

[32] K. J. Jeon, H. Choi, C. Lee, and S. S. Han, “Automatic Diagnosis of True Proximity Between the Mandibular Canal and the Third Molar on Panoramic Radiographs Using Deep Learning,” Scientific Reports 13, no. 1 (2023): 22022.

[33] E. Choi, S. Lee, E. Jeong, et al., “Artificial Intelligence in Positioning Between Mandibular Third Molar and Inferior Alveolar Nerve on Panoramic Radiography,” Scientific Reports 12, no. 1 (2022): 2456.

[34] E. D. Costa, P. D. Peyneau, M. A. Visconti, K. L. Devito, G. M. B. Ambrosano, and F. S. Verner, “Double Mandibular Canal and Triple Mental Foramina: Detection of Multiple Anatomical Variations in a Single Patient,” General Dentistry 67, no. 5 (2019): 46–49.

[35] K. Orhan, S. Aksoy, B. Bilecenoglu, B. U. Sakul, and C. S. Paksoy, “Evaluation of Bifid Mandibular Canals With Cone-Beam Computed Tomography in a Turkish Adult Population: A Retrospective Study,” Surgical and Radiologic Anatomy 33, no. 6 (2011): 501–507.

[36] T. von Arx, S. Bolt, and M. M. Bornstein, “Neurosensory Disturbances After Apical Surgery of Mandibular Premolars and Molars: A Retrospective Analysis and Case-Control Study,” European Endodontic Journal 6, no. 3 (2021): 247–253.

[37] D. Zhan, F. Wang, D. Zeng, et al., “Development of a Radiomic Model to Detect the Retromolar Canal on Panoramic Radiographs,” Discovery Medicine 37, no. 193 (2025): 383–393.

[38] N. Jha, K. Lee, and Y. J. Kim, “Diagnosis of Temporomandibular Disorders Using Artificial Intelligence Technologies: A Systematic Review and Meta-Analysis,” PLoS One 17, no. 8 (2022): e0272715.

[39] Z. Wu, X. Yu, F. Wang, and C. Xu, “Application of Artificial Intelligence in Dental Implant Prognosis: A Scoping Review,” Journal of Dentistry 144 (2024): 104924.

[40] B. Öztürk, A. Altındağ, Ö. Çelik, I. S. Bayrakdar, and K. Orhan, “Pulp-Stone Detection in Panoramics Using Deep Learning: A Multi-Institutional Study,” International Dental Journal 74 (2024): S40–S41.

[41] M. Naitoh, Y. Hiraiwa, H. Aimiya, and E. Ariji, “Observation of Bifid Mandibular Canal Using Cone-Beam Computerized Tomography,” International Journal of Oral & Maxillofacial Implants 24, no. 1 (2009): 155–159.

[42] A. Auluck, A. Ahsan, K. M. Pai, and C. Shetty, “Anatomical Variations in Developing Mandibular Nerve Canal: A Report of Three Cases,” Neuroanatomy 4 (2005): 28–30.

[43] K. Mizbah, N. Gerlach, T. J. Maal, S. J. Bergé, and G. J. Meijer, “The Clinical Relevance of Bifid and Trifid Mandibular Canals,” Oral and Maxillofacial Surgery 16, no. 1 (2012): 147–151.

[44] K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” Presented at: 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR); 2016.

[45] D. P. Kingma and J. Ba, “Adam: A Method for Stochastic Optimization,” in 2014 International Conference for Learning Representations (ICLR) (2014).

[46] I. Loshchilov and F. Hutter, “Sgdr: Stochastic Gradient Descent With Warm Restarts,” in 2017 International Conference for Learning Representations (ICLR) (2017).

RIGHTS & PERMISSIONS

2025 The Author(s). MedComm - Future Medicine published by John Wiley & Sons Australia, Ltd on behalf of Sichuan International Medical Exchange & Promotion Association (SCIMEA).