"Model for Detecting Bruxism Based on Images of Tongues with Dental Marks Using Computer Vision"

Abstract

"Bruxism is a complex functional disorder that has garnered increasing concern in the field of dentistry due to its high prevalence and significant clinical impact when undiagnosed. This study proposes a robust model using advanced computer vision techniques for the early detection of bruxism through the analysis of dental marks in tongue images. The methodology was developed in four phases: data acquisition (using a set of tongue images with labeled conditions of bruxism), preprocessing (normalization and resizing), application of deep learning models (Vision Transformers, EfficientNet, Swin Transformers, and others), and evaluation using key metrics such as accuracy, precision, recall, and F1-score. The results indicate that the optimized models, especially EfficientNet, outperformed their initial performance, with significant improvements after hyperparameter tuning. The confusion matrix demonstrated the models’ high ability to classify different bruxism conditions accurately, with a perfect ROC curve (AUC?=?0.997) confirming their discriminative power. These findings highlight the potential of optimized deep learning models to support early bruxism diagnosis, paving the way for their practical application in clinical settings, improving diagnostic accuracy and facilitating timely interventions to enhance patient care. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026."