Enhancing the Image-Based Gingivitis Disease Detection System Using Deep Learning Approach

Abstract

Gingivitis is one of the most prevalent oral health conditions worldwide and represents an early stage of periodontal disease. If not detected and treated promptly, gingivitis may progress into periodontitis, leading to severe gum damage, tooth loss, and systemic health complications. Traditional diagnosis of gingivitis requires clinical examination by dental professionals, which is often inaccessible in remote or underserved regions. This paper presentsEnhancing The Image Based Gingivitis Disease Detection System Using Deep Learning Approach , an intelligent web-based system for automated detection of gingivitis using deep learning techniques. The system employs a fine-tuned EfficientNet-B1 convolutional neural network architecture for binary classification of gum health conditions into two categories: healthy gums and gingivitis. The model was trained on a curated dataset of 2,800 oral images, utilizing data augmentation strategies including rotation, zooming, brightness adjustment, and horizontal flipping. The model achieved a test accuracy of 90% and a ROC-AUC score of 0.956, demonstrating reliable discriminative capability. The trained model is integrated into a Flask-based REST API backend paired with an HTML/JavaScript frontend, enabling users to upload gum images and receive real-time classification results with confidence scores. Gingivitis Disease Detection System provides a practical and efficient solution for preliminary gum health screening and clinical decision support.

Introduction

This research presents GumCheck, an AI-powered gingivitis detection system that uses EfficientNet-B1, a deep convolutional neural network (CNN), to automatically identify gingivitis from gum images. Gingivitis is a common inflammatory gum disease caused by bacterial plaque and, if left untreated, can progress to periodontitis, which is associated with serious systemic diseases such as cardiovascular disease and diabetes. Since early-stage gingivitis often exhibits mild symptoms and requires professional dental examination for diagnosis, the proposed system aims to provide an accessible and efficient tool for early detection.

The study leverages transfer learning with the EfficientNet-B1 architecture, initialized with ImageNet pretrained weights and fine-tuned on a curated dataset of 2,800 oral images comprising healthy gums and gingivitis cases. Images are preprocessed through resizing, normalization, and data augmentation techniques such as rotation, zooming, brightness adjustment, and horizontal flipping to improve model generalization. The final network includes global average pooling, dense layers, dropout regularization, and a sigmoid output layer for binary classification.

GumCheck is deployed as a Flask-based web application featuring a user-friendly HTML, CSS, and JavaScript interface. Users can upload gum images through a web browser, after which the backend preprocesses the image, performs inference using the trained EfficientNet-B1 model, and returns the predicted class along with probability and confidence scores via a REST API. The system supports real-time diagnosis and is designed for deployment on cloud platforms using Docker and production web servers.

Experimental evaluation on a 420-image test set demonstrates strong performance, achieving 90.0% accuracy, 0.90 precision, 0.90 recall, 0.90 F1-score, and a ROC-AUC score of 0.956. Compared with previous approaches based on VGG-16 (84.5%), ResNet-50 (87.3%), and MobileNet-V2 (85.0%), GumCheck achieves the highest classification accuracy. The improvement is attributed to EfficientNet-B1's efficient architecture, transfer learning, comprehensive data augmentation, and the larger, well-curated dataset.

Conclusion

This paper has presented Gingivitis Disease Detection System, a deep learning-based web application for automated detection of gingivitis from oral images. The system employs EfficientNet-B1, fine-tuned using transfer learning on a curated dataset of 2,800 gum images, and achieves 90% classification accuracy and a ROC-AUC score of 0.956 on a held-out test set. The end-to-end architecture integrates a Flask REST API backend with an HTML/JavaScript frontend, enabling real-time image classification with confidence reporting. The results demonstrate that GumCheck reliably distinguishes between healthy gums and gingivitis, outperforming related models in the literature. While the system is intended as a decision support tool rather than a clinical diagnostic replacement, it holds significant potential for improving accessibility to preliminary oral health screening in underserved and remote populations. Future work will focus on multi-class severity classification, Grad-CAM visualization for explainability, mobile deployment, and dataset expansion through clinical partnerships. GumCheck represents a meaningful step toward AI-augmented oral healthcare, demonstrating how deep learning and web technologies can be combined to create practical, accessible, and clinically relevant diagnostic tools.

References

[1] Tan, M., & Le, Q. V. (2019). EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks. Proceedings of the 36th International Conference on Machine Learning (ICML 2019), PMLR 97, 6105–6114. [2] Rajpurkar, P., Irvin, J., Ball, R. L., et al. (2017). CheXNet: Radiologist-Level Pneumonia Detection on Chest X-Rays with Deep Learning. arXiv preprint arXiv:1711.05225. [3] Esteva, A., Kuprel, B., Novoa, R. A., et al. (2017). Dermatologist-level classification of skin cancer with deep neural networks. Nature, 542(7639), 115–118. [4] Jiang, L., Chen, D., Cao, C., et al. (2021). A two-stage deep learning architecture for radiographic staging of periodontal bone loss. NPJ Digital Medicine, 4(1), 97. [5] Lee, J. H., Kim, D. H., Jeong, S. N., & Choi, S. H. (2020). Detection and diagnosis of dental caries using a deep learning-based convolutional neural network algorithm. Journal of Dentistry, 77, 106–111. [6] Jain, M., Nainani, A., Gupta, P., et al. (2022). Oral lesion detection and classification using transfer learning with data augmentation. Journal of Biomedical Informatics, 128, 104041. [7] Kim, Y. J., Jang, H., Lee, K., et al. (2020). A deep learning approach for detecting colorectal cancer via colonoscopy images. Diagnostics, 10(12), 1012. [8] Nguyen, T., Do, T., Dao, V., et al. (2021). Multi-class classification of oral lesions using residual neural networks. Proceedings of the International Conference on Biomedical Engineering (ICBME 2021), 214–219. [9] Prasetyo, E., Suciati, N., & Fatichah, C. (2022). Gingivitis detection on mobile device using lightweight convolutional neural network. Electronics, 11(8), 1237. [10] Joshi, K., Arora, A., Bhatt, H., et al. (2021). Mobile-based AI diagnosis for oral health in rural telemedicine settings. Journal of Telemedicine and Telecare, 27(4), 221–230. [11] Ozden, F. O., Ozgonenel, O., Ozden, B., & Aydogdu, A. (2020). Diagnosis of periodontal diseases using radial basis function neural networks. Turkish Journal of Electrical Engineering and Computer Sciences, 23(2), 466–476. [12] Tuzoff, D. V., Tuzova, L. N., Bornstein, M. M., et al. (2019). Tooth detection and numbering in panoramic radiographs using convolutional neural networks. Dentomaxillofacial Radiology, 48(4), 20180051. [13] Chollet, F. (2017). Xception: Deep Learning with Depthwise Separable Convolutions. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2017), 1251–1258. [14] Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep Learning. MIT Press, Cambridge, MA. [15] Ronneberger, O., Fischer, P., & Brox, T. (2015). U-Net: Convolutional Networks for Biomedical Image Segmentation. Medical Image Computing and Computer-Assisted Intervention (MICCAI 2015), Lecture Notes in Computer Science, 9351, 234–241.

Copyright

Copyright © 2026 Naredla Yuvaraj, Kappala Tharun Kumar, Ravula Shiva Prasad, Pallerla Arun Kumar, Chandrugonda Malleswar Rao. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.