Deep Learning-Based Dental Image Analysis Using Grad Cam Convolutional Network
Deep learning; Explainable artificial intelligence (XAI); Grad-CAM; Instance segmentation; Panoramic dental radiographs.
AuthorsAbstractPanoramic radiographs have been regularly applied in the field of prosthodontic and restorative dentistry to aid in treatment planning, tooth morphology, edentulous space evaluation, and preliminary screening of the implant site. Nevertheless, panoramic images may not be easily interpreted because of anatomy overlap, distortion, and varying quality of images. This paper provides a deep learning architecture that can be explained and used to aid the radiographic evaluation of prosthodontics by means of automated localization of teeth and interpretable visualisation. It applied a multi-stage pipeline, which involved dataset validation, COCO-based annotation auditing, instance segmentation with a Mask R-CNN backbone based on ResNet-FPN, and incorporation of explainable artificial intelligence (XAI) methods. Grad-CAM, occlusion sensitivity mapping, and mask confidence visualisation were automatically used to segment and analyse tooth regions to get transparent decision-support outputs. Qualitative data showed that there was anatomical localization and activation of teeth in patterns that were consistent with morphologically relevant structures. Despite the fact that quantitative metrics of segmentation were affected by rigid confidence levels, explainability analysis showed that model attention was mainly focused on tooth anatomy instead of background artefact. The suggested framework offers a reproducible and interpretable base of AI-assisted panoramic radiograph interpretation, which has possible applications in the field of prosthodontic planning, restorative assessment, and implant-oriented screening procedures. 1.Introduction Panoramic dental radiography has become a common tool in the field of prosthodontics and restorative dentistry, to assess the condition of the dentition, alveolar bone support, tooth morphology, edentulous spaces, and the condition of the maxillofacial region before the application of the restorative interventions. It is also a vital part of treatment planning of fixed prostheses, removable partial dentures, full-mouth rehabilitation, and implant-supported restorations. Nevertheless, panoramic radiographs are difficult to interpret because of structural overlap, distortion artifacts, differences in contrast, and changes in patient positioning. Such factors can decrease diagnostic consistency especially when several teeth and anatomical landmarks should be evaluated at the same time. Therefore, automated image analysis systems are also becoming a more studied decision-support tool to make workflows in restorative activities more efficient and reliable. Deep learning, especially convolutional neural networks (CNNs) have shown great potential in dental image detection as they make it possible to automatically detect, segment and classify anatomical structures and pathological conditions. CNN-based has been effectively used in detecting dental caries and interpreting panoramic images, and Grad-CAM visualization has been used to facilitate clear clinical validation [1]. Deep learning has also emerged as a powerful approach to dental and maxillofacial imaging studies, with positive outcomes in segmentation, detection, and classification tasks [2]. These advancements demonstrate the practicability of using deep learning to panoramic radiographs to localize and analyze the structure of the teeth automatically. Although such deep learning models have high predictive performance, most of them are used as black-box systems, producing results without the radiographic features that determine decisions. This Uninterpretability has restricted clinical acceptability, especially in the area of prosthodontics and restorative dentistry, where the treatment planning involves • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • ejprd.org - Published by Riset Publishing Services LLC
Copyright ©2026 by Riset Publishing Services LLC
/ P38