Options
Caries detection in dental imaging using vision transformer and explainable AI
Date Issued
2025
Citation
Shabbir, I., Azhar, M., Shafqat Ali, M., Chan, K. L., Wattoo, W. A., & Iqbal, A. (2025 Aug 26). Caries detection in dental imaging using bision transformer and explainable AI. AI-SI 2025 - IEEE International Conference on Artificial Intelligence for Sustainable Innovation: Shaping the Future with Intelligent Solutions, Seri Pacific Hotel Kuala Lumpur, Kuala Lumpur.
Type
Conference Paper
Abstract
Dental caries, commonly known as tooth decay or cavities, have long posed a significant challenge to oral health. Early identification is essential for effective treatment and for
preventing further damage to teeth. Traditionally, dentists relied on visual examinations to diagnose cavities by looking for obvious signs of decay. However, due to the increasing
patient-to-dentist ratio and the advancement of artificial intelligence, deep learning techniques have gained prominence in recent years. Despite their potential, the black-box
nature of these deep learning models can lead to ambiguity in their decision-making processes. To tackle this issue, a study was conducted using explainable Al in human-centric medical applications to identify cavities in dental images and provide clear visualizations for diagnosis. Our approach integrates Vision Transformer (ViT) with Local Interpretable Model-Agnostic Explanations (LIME), where LIME is utilized to visually elucidate the classifications made by the ViT model. The Vision Transformer effectively extracts relevant features from dental images, while LIME offers interpretable visual explanations for the Al model's decisions, helping dentists understand the factors influencing their diagnoses. Experimental results demonstrate the efficacy of our approach in cavity detection, facilitating early intervention and ultimately improving oral health outcomes through
explainable Al.
Availability at HKSYU Library
