Machine Learning and Deep Learning Applications in Oral Surgery and Dental Sciences
European Journal of Prosthodontics and Restorative Dentistry (2026) 34(1s), 157-165
KeywordsMachine learning, Dental radiographs, Tooth classification, Panoramic imaging, Oral surgery European Journal of Prosthodontics and Restorative Dentistry (2026) 34(1s), 157-165
AuthorsAbstractObjectives; Cigarette smoking is a clinically relevant risk factor in dentistry because it may influence systemic metabolism, tissue healing, Artificial intelligence has growing relevance in oral surgery and dental sciences, particularly for radiographic interpretation, tooth localization, and image-based decision support. Interpretable machine learning approaches may provide practical value where structured anatomical features are available from annotated dental images. The present study evaluated the application of machine learning for automated tooth classification and anatomical analysis using annotated panoramic dental radiographs. A secondary dataset of annotated panoramic dental radiographs was analyzed. Pascal VOC annotation files were processed to extract tooth-class labels, image dimensions, bounding-box coordinates, and derived geometric features. Descriptive and morphological analyses were performed, followed by supervised multiclass classification using a Random Forest model. Model performance was evaluated using accuracy, precision, recall, F1-score, confusion matrix analysis, feature-importance assessment, and five-fold cross-validation. A total of 14,227 annotated tooth regions from 585 radiographs and 16 tooth classes were analyzed. The model achieved a test-set accuracy of 90.62%, with weighted precision, recall, and F1-score values above 90%. Five-fold cross validation showed stable performance, with a mean accuracy of 90.33% and a low standard deviation of 0.0032. Feature-importance analysis indicated that horizontal tooth-region coordinates were the strongest predictors of classification performance. The findings support the use of lightweight and interpretable machine learning models for dental image analysis, radiographic assessment, and decision-support applications in oral surgery and dental sciences. 1. INTRODUCTION Artificial Intelligence (AI) is revolutionizing healthcare, with its ability to process large volumes of clinical data, assist in diagnoses, and streamline treatment processes. Among the several subfields of AI, two of the important fields are machine learning (ML) and deep learning (DL), which have been really helpful in medical imaging, disease prediction and automated healthcare analysis [1]. With the growing influence of AI on radiological processes, intelligent systems have been rapidly evolving to identify intricate anatomical patterns and aid in clinical interpretation. As computational image analysis has been developed, the applications of AI based image analysis in healthcare imaging sciences have been enhanced even further [2]. There has been a significant digital transformation in dental and oral healthcare, with the use of digital radiographic systems now being widespread. A panoramic dental radiograph is useful for the overall overview of the maxillofacial structures and is frequently used in oral surgery, oral implant planning, orthodontics and diagnostic dentistry [3]. With the increasing accessibility of digital dental imaging, opportunities have arisen to leverage AI techniques for image analysis that can support clinicians in image interpretation and anatomical localization. The advent of the latest •••••••••••••••••••••••••••••••• ejprd.org- Published by Riset Publishing Services LLC.
EJPRD
Copyright © 2026 by Riset Publishing Services LLC