An Intelligent Method for Detecting and Classifying Dental Caries from Dental Radiographic Images

In recent years, dental image processing has become an essential tool in the early diagnosis and management of dental diseases, particularly dental caries. This technology addresses inherent limitations in traditional dental radiographs, such as low contrast and overlapping anatomical structures. However, despite technological progress, the accurate detection of dental caries remains a challenging task primarily due to the variability and non-uniformity of dental X-ray images. Most existing computer-aided diagnostic (CAD) systems rely heavily on supervised learning models that require large, annotated datasets. These models often perform sub optimally when confronted with images that differ significantly from the training data, leading to diagnostic inaccuracies. In this study, we propose an innovative method for tooth segmentation and caries detection from a diverse set of dental X-rays using an unsupervised learning approach. Unlike conventional systems, the proposed method employs a diagnostic protocol inspired by clinical dental evaluations, enabling the system to assess carious lesions relative to the structure and features of each individual image rather than relying on fixed detectors. Experimental results demonstrate that our method achieves a diagnostic accuracy of ۹۶%, outperforming current supervised approaches. These findings highlight the robustness and adaptability of the proposed unsupervised framework, making it a promising solution for real-world dental diagnostic applications.

Keywords:

x-Ray images , Unsupervised Learning , Dental Caries