AI-Enabled Clinical Decision Support for Oral Disease Detection

Abstract

Oral health is a critical component of global public health, with dental diseases such as caries, periodontal conditions, periapical lesions, and impacted teeth affecting millions of individuals annually. These conditions not only lead to pain, infection, and tooth loss but are also linked to systemic health issues, including cardiovascular disease, diabetes and respiratory infections. Given the significant burden of oral diseases, early and accurate diagnosis is essential to prevent complications and improve patient outcomes. This study explores the feasibility of AI-based deep learning models, specifically YOLO architectures, for the automated detection and classification of dental diseases using panoramic radiographs. A comprehensive dataset containing 50 plus annotated dental disease categories was used to train and evaluate YOLOv5, YOLOv8, and YOLO-NAS models. To optimize model performance, a systematic data preprocessing pipeline incorporating Cropping and Region of Interest Selection, Conversion to Grayscale, Resolution Standardization, Dataset Splitting and Non-Enhanced Datasets was implemented. Additionally, data augmentation techniques were employed to enhance model generalization, including Geometric Transformations, Rotation, Flipping, Grid Masking and to further improve model efficiency. Performance metrics, including precision, recall, and accuracy, were used to evaluate model effectiveness. Among the tested models, YOLOv5 outperformed YOLOv8 and YOLO-NAS, achieving a precision of 95.7%, recall of 96%, and accuracy of 96.5%, making it the most effective model for dental disease detection. The proposed AI-powered dental diagnostic framework demonstrated high accuracy, real-time detection capabilities, and adaptability across multiple disease categories, positioning it as a viable tool for clinical applications. Beyond clinical practice, AI-driven dental diagnostics offer significant public health benefits, particularly in rural and underserved areas where access to specialized dental care is limited. The ability to automate radiographic analysis facilitates large-scale screening programs, enabling early disease detection, cost-effective intervention strategies, and improved healthcare accessibility. Additionally, AI-based diagnostics contribute to reducing consultation times, and minimizing reliance on specialist interpretations. Given the strong association between oral health and systemic diseases, early AI-powered diagnosis can also help mitigate broader health risks such as diabetes complications and cardiovascular conditions. By integrating AI into routine dental practice and public health initiatives, this study underscores the transformative potential of deep learning in enhancing diagnostic precision and improving overall healthcare outcomes. These findings establish hyperparameter tuned YOLOv5 as the most effective model for AI-driven dental disease detection. This research paves the way for next-generation AI-assisted dental diagnostic platforms, fostering early intervention, equitable healthcare access, and improved global oral health outcomes.