MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202641058546 A) filed by Agni College Of Technology on May 08, 2026, for Optimized Deep Learning Pipeline For High Precision Dental Cavity Classification Using Sota Computer Vision.
Inventors include Manoranjitham. S; Vaheetha Banu S; Dhanush R; and Mohamed Aamer M A.
The application for the patent was published on July 31, 2026, under issue no. 31/2026.
Abstract: Dental caries remains one of the most common global health problems, affecting over 3.5 billion people worldwide. Early detection is crucial, yet traditional diagnostic methods are often subjective, require radiation, and may not be accessible in many regions. Delayed diagnosis can lead to severe complications and expensive treatments. To address these challenges, this project proposes an optimized deep learning pipeline for accurate and automated dental cavity detection and classification using intraoral images. The system leverages advanced State-of-the-Art (SOTA) computer vision techniques by integrating three key deep learning models. A Generative Adversarial Network (GAN) is used for data augmentation and image enhancement, helping to overcome the limitation of small medical datasets. A U-Net architecture is applied for precise segmentation of teeth and lesion boundaries, ensuring accurate identification of affected regions. Finally, a YOLOv11 model is utilized for real-time detection and classification of cavities into three stages: incipient, moderate, and severe. Based on the detected stage, the system provides appropriate treatment recommendations, ranging from preventive care for early-stage cavities to clinical intervention for advanced cases. Inspired by real-time monitoring systems such as automated traffic detection, the proposed pipeline delivers fast and near-instantaneous results. The model achieves high performance with a mean Average Precision (mAP) of 94.2% and classification accuracy of 89.7%. It is deployed as a user-friendly web application using Django, allowing users to upload images, view results, and receive clinical decision support. This system offers a non-invasive, radiation-free, and accessible solution for dental screening. By enabling early detection and timely intervention, it helps reduce treatment costs, improve patient outcomes, and expand access to dental care, especially inunderserved communities.
Disclaimer: Curated by HT Syndication.
