Deep Learning Approaches for Dental Plaque Detection: A Scoping Review of Methods, Diagnostic Performance, and Clinical Applications

Abstract Dental plaque is a primary etiological factor in dental caries and periodontal diseases, making its accurate detection essential for preventive dentistry and oral health monitoring. Traditional plaque evaluation methods are subjective and prone to examiner variability. Recent advances in artificial intelligence, particularly deep learning, have enabled automated plaque detection using digital dental images; however, existing studies vary widely in methodology and reporting. The review aimed to map the current literature on deep learning approaches for dental plaque detection, summarize reported diagnostic performance metrics, and examine described clinical and public health applications. This review was conducted in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews guidelines. Multiple electronic databases were searched for studies published between 2020 and 2025 that applied deep learning models for dental plaque detection, segmentation, or classification using image-based dental data. Data charting was performed descriptively to summarize study characteristics, model architectures, diagnostic tasks, performance measures, and clinical applications. Fourteen studies met the inclusion criteria. Most studies employed convolutional neural network-based architectures, including U-Net, DeepLabV3+, YOLO, and multitask learning frameworks, using intraoral red, green, and blue images as input. Diagnostic performance varied across studies, with accuracy and specificity generally reported at high levels, whereas sensitivity values ranged from approximately 65% to 84%. Favorable segmentation performance was frequently reported using metrics such as intersection over union and dice coefficients Id entified ap lications inc luded aut omated pla que ind exing, cha irside decision support, patient education, and smartphone-based oral hygiene monitoring. Variability was observed in datasets, annotation protocols, and performance reporting. The literature reveals increasing research interest in deep learning-based dental plaque detection and suggests promising diagnostic capability. However, methodological heterogeneity, limited external validation, and a lack of large-scale clinical studies remain challenges. Further research is needed to support clinical translation.