Year: 2025
Type: article
Abstract: Dental caries is a disease that is common worldwide.Dental caries remains one of the most prevalent oral diseases globally, necessitating accurate and early detection to enable timely interventions and preserve tooth structure.This study presents a lightweight and task-specific Convolutional Neural Network (CNN) architecture, called DentalNet-Lite, designed for the multiclass classification of 1200 RGB intraoral images into three diagnostic categories: normal tooth, early dental caries, and advanced dental caries.The model DentalNet-Lite is a custom CNN designed for the automated detection of dental caries.The model was trained and evaluated, benchmarked against five pretrained CNNs: MobileNetV2, DenseNet121, ResNet50V2, Xception, and InceptionResNetV2.employing accuracy, precision, recall, and F1-score as evaluation criteria.DentalNet-Lite achieved a test accuracy of 99.07%, exceeding all competing approaches while maintaining low computing complexity, hence demonstrating its suitability for real-time, resource-limited clinical applications.
Authors: Ali Qasim Ali, Hanan M. Amer, Mohamed M. Abdelsalam, Mohamed S. Saraya
Institution: Mansoura University
Language: English
Subfield: Oral Surgery
Field: Dentistry
Domain: Health Sciences
Open Access status: bronze
OpenAlex ID: W4415347545
FWCI:
Cites:
Cited by:
Related to: