Dental caries detection and treatment cost prediction using YOLOv11

Abstract

Dental caries is one of the most prevalent oral diseases, progressively damaging tooth structure and often leading to significant treatment costs. Variations in dental service fees across clinics can become a financial barrier, discouraging timely and appropriate care. This study introduces an artificial intelligence (AI)-based framework that utilises smartphone camera images to detect dental caries and predict treatment costs. A total of 1,200 images of carious and normal teeth were collected from dental clinics in Denpasar, Bali, Indonesia, and classified by three dental experts. Data augmentation expanded the dataset twentyfold to 23,060 images to address variation and class imbalance. The you only look once version 11 (YOLOv11) deep learning algorithm was employed for caries detection, and its performance was evaluated using mean average precision (mAP), precision, and recall metrics. The model demonstrated high accuracy, achieving an mAP of 96.1%, a precision of 95.5%, and a recall of 93.0%. This study provides the first integration of YOLOv11 with RGB-intensity-based cost prediction in digital dentistry. The proposed system offers a fast, accessible, and cost-efficient approach for early caries detection and treatment cost estimation. These findings highlight its potential to support real-time, AI-assisted preventive dentistry and contribute to more equitable access to oral healthcare.