Segmentation of Pediatric Carious Lesions in Panoramic Dental X-Rays via Deep Learning Model with Atrous Spatial Pyramid Pooling: Explainable Artificial Intelligence Study
Main Article Content
Abstract
Pediatric tooth decay is one of the common non-infectious ailments globally; however, its early diagnosis through image analysis is limited due to the lack of objective evaluation and poor boundary identification. In this study, we evaluate DeepLabV3+, a semantic segmentation model conFig.d with a ResNet-34 encoder in identifying pediatric dental caries from panoramic X-rays. Through training on a dataset of 100 children's panoramic X-rays consisting of 968 manually segmented tooth decay lesions using a five-fold cross-validation technique, our method demonstrated an average Intersection over Union of 0.835 ± 0.004, a pixel accuracy of 0.904 ± 0.003, and a Dice similarity of 0.846 ± 0.004. Ablation experiments demonstrate that the parallel ASPP model gives rise to 14.5% relative performance improvement from the baseline encoder, whereas low-level encoder skip connections provide an extra 7.1% relative improvement in the boundary refinement process. The explainable artificial intelligence-based analysis using the gradient-weighted class activation map shows that the proposed model concentrates more spatially on the actual pathologic margin by 23% compared to the baseline model, thereby improving clinical accuracy to 0.678 from 0.612. Evaluations based on screening performance criteria provide 98.1% specificity and 96.6% negative predictive value, thus ensuring solid applicability for public health practice as an automated tool for rule-out triage under constrained diagnostics. While moderate sensitivity of 54.3% calls for persistent clinical supervision and does not allow for complete automation, the open-source framework ensures excellent repeatability for precision dental informatics.
Article Details
This work is licensed under a Creative Commons Attribution 4.0 International License.
References
[1] N. B. Pitts et al., “Dental caries,” Nat. Rev. Dis. Prim., vol. 3, no. 1, p. 17030, May 2017, doi: 10.1038/nrdp.2017.30.
[2] N. J. Kassebaum, E. Bernabé, M. Dahiya, B. Bhandari, C. J. L. Murray, and W. Marcenes, “Global Burden of Untreated Caries,” J. Dent. Res., vol. 94, no. 5, pp. 650–658, May 2015, doi: 10.1177/0022034515573272.
[3] M. A. Peres et al., “Oral diseases: a global public health challenge,” Lancet, vol. 394, no. 10194, pp. 249–260, Jul. 2019, doi: 10.1016/S0140-6736(19)31146-8.
[4] M. Fontana and D. T. Zero, “Assessing patients’ caries risk,” J. Am. Dent. Assoc., vol. 137, no. 9, pp. 1231–1239, Sep. 2006, doi: 10.14219/jada.archive.2006.0380.
[5] A. Wenzel, “Bitewing and Digital Bitewing Radiography for Detection of Caries Lesions,” J. Dent. Res., vol. 83, no. 1_suppl, pp. 72–75, Jul. 2004, doi: 10.1177/154405910408301s14.
[6] S. C. White and M. J. Pharoah, Oral Radiology - E-Book: Principles and Interpretation. Elsevier Health Sciences, 2014. [Online]. Available: https://books.google.co.id/books?id=V7PwAwAAQBAJ
[7] A. S. P. S. da Rocha, S. M. de Almeida, F. N. Bóscolo, and F. Haiter Neto, “Interexaminer agreement in caries radiographic diagnosis by conventional and digital radiographs,” J. Appl. Oral Sci., vol. 13, no. 4, pp. 329–333, Dec. 2005, doi: 10.1590/S1678-77572005000400003.
[8] J.-H. Lee, D.-H. Kim, S.-N. Jeong, and S.-H. Choi, “Detection and diagnosis of dental caries using a deep learning-based convolutional neural network algorithm,” J. Dent., vol. 77, pp. 106–111, Oct. 2018, doi: 10.1016/j.jdent.2018.07.015.
[9] J. Krois et al., “Deep Learning for the Radiographic Detection of Periodontal Bone Loss,” Sci. Rep., vol. 9, no. 1, p. 8495, Jun. 2019, doi: 10.1038/s41598-019-44839-3.
[10] O. Ronneberger, P. Fischer, and T. Brox, “U-net: Convolutional Networks for Biomedical Image Segmentation,” in International Conference on Medical image computing and computer-assisted intervention, 2015, pp. 234–241.
[11] A. G. Cantu et al., “Detecting caries lesions of different radiographic extension on bitewings using deep learning,” J. Dent., vol. 100, p. 103425, Sep. 2020, doi: 10.1016/j.jdent.2020.103425.
[12] L.-C. Chen, Y. Zhu, G. Papandreou, F. Schroff, and H. Adam, “Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation,” V. Ferrari, M. Hebert, C. Sminchisescu, and Y. Weiss, Eds., Cham: Springer International Publishing, 2018, pp. 833–851. doi: 10.1007/978-3-030-01234-2_49.
[13] J.-H. Lee, S.-S. Han, Y. H. Kim, C. Lee, and I. Kim, “Application of a fully deep convolutional neural network to the automation of tooth segmentation on panoramic radiographs,” Oral Surg. Oral Med. Oral Pathol. Oral Radiol., vol. 129, no. 6, pp. 635–642, Jun. 2020, doi: 10.1016/j.oooo.2019.11.007.
[14] S. Vinayahalingam, T. Xi, S. Bergé, T. Maal, and G. de Jong, “Automated detection of third molars and mandibular nerve by deep learning,” Sci. Rep., vol. 9, no. 1, p. 9007, Jun. 2019, doi: 10.1038/s41598-019-45487-3.
[15] Y. Zhang et al., “Children’s Dental Panoramic Radiographs Dataset for Caries Segmentation and Dental Disease Detection,” 2023. doi: 10.6084/m9.figshare.c.6317013.v1.
[16] J. Deng et al., “ImageNet: A large-scale hierarchical image database,” in 2009 IEEE Conference on Computer Vision and Pattern Recognition, IEEE, Jun. 2009, pp. 248–255. doi: 10.1109/CVPR.2009.5206848.
[17] A. Buslaev, V. I. Iglovikov, E. Khvedchenya, A. Parinov, M. Druzhinin, and A. A. Kalinin, “Albumentations: Fast and Flexible Image Augmentations,” Information, vol. 11, no. 2, p. 125, Feb. 2020, doi: 10.3390/info11020125.
[18] P. Iakubovskii, “Segmentation Models Pytorch,” GitHub Repos., 2019, [Online]. Available: https://github.com/qubvel/segmentation_models.pytorch
[19] E. Asci et al., “A Deep Learning Approach to Automatic Tooth Caries Segmentation in Panoramic Radiographs of Children in Primary Dentition, Mixed Dentition, and Permanent Dentition,” Children, vol. 11, no. 6, p. 690, Jun. 2024, doi: 10.3390/children11060690.
[20] A. A. Al Tuwirqi, “Eye-Tracking Technology in Dentistry: A Review of Literature,” Cureus, Feb. 2024, doi: 10.7759/cureus.55105.
[21] R. R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D. Parikh, and D. Batra, “Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization,” in 2017 IEEE International Conference on Computer Vision (ICCV), IEEE, Oct. 2017, pp. 618–626. doi: 10.1109/ICCV.2017.74.
[22] L. Maier-Hein et al., “Metrics reloaded: recommendations for image analysis validation,” Nat. Methods, vol. 21, no. 2, pp. 195–212, Feb. 2024, doi: 10.1038/s41592-023-02151-z.
[23] A. Sabarudin and Y. J. Tiau, “Image quality assessment in panoramic dental radiography: a comparative study between conventional and digital systems,” Quant. Imaging Med. Surg., vol. 3, no. 1, 2013, doi: 10.3978/j.issn.2223-4292.2013.02.07.
[24] H. Zhao, J. Shi, X. Qi, X. Wang, and J. Jia, “Pyramid Scene Parsing Network,” 2017 IEEE Conf. Comput. Vis. Pattern Recognit., pp. 6230–6239, 2016, [Online]. Available: https://api.semanticscholar.org/CorpusID:5299559
[25] T.-Y. Lin, P. Goyal, R. Girshick, K. He, and P. Dollar, “Focal Loss for Dense Object Detection,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 42, no. 2, pp. 318–327, Feb. 2020, doi: 10.1109/TPAMI.2018.2858826.
[26] G. Jocher, Q. Jing, M. Liu, S. Lyu, F. C. Akyon, and M. E. Kalfaoglu, “Ultralytics YOLO26: Unified Real-Time End-to-End Vision Models,” 2023, doi: 10.48550/arXiv.2606.03748.
[27] G. Wang, W. Li, S. Ourselin, and T. Vercauteren, “Automatic Brain Tumor Segmentation Based on Cascaded Convolutional Neural Networks With Uncertainty Estimation,” Front. Comput. Neurosci., vol. 13, Aug. 2019, doi: 10.3389/fncom.2019.00056.
[28] European Commission, “Proposal for a Regulation laying down harmonised rules on artificial intelligence,” Policy Regul. Propos., no. April, p. 2025, 2021, [Online]. Available: https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:52021PC0206
[29] C. Chen, Q. Dou, H. Chen, J. Qin, and P.-A. Heng, “Synergistic Image and Feature Adaptation: Towards Cross-Modality Domain Adaptation for Medical Image Segmentation,” Proc. AAAI Conf. Artif. Intell., vol. 33, no. 01, pp. 865–872, Jul. 2019, doi: 10.1609/aaai.v33i01.3301865.
[30] S. Y. L. Kwan, P. E. Petersen, C. M. Pine, and A. Borutta, “Health-promoting schools: an opportunity for oral health promotion,” Bull. World Health Organ., vol. 83, no. 9, pp. 677–85, Sep. 2005, [Online]. Available: http://www.ncbi.nlm.nih.gov/pubmed/16211159
[31] Z. Z. Qin et al., “Using artificial intelligence to read chest radiographs for tuberculosis detection: A multi-site evaluation of the diagnostic accuracy of three deep learning systems,” Sci. Rep., vol. 9, no. 1, p. 15000, Oct. 2019, doi: 10.1038/s41598-019-51503-3.