A Dataset of Panoramic Radiographs for Two-Stage Dental Caries Detection and Classification
Description
This dataset contains 1,427 de-identified digital panoramic radiographs (orthopantomograms, OPGs) curated for the development and evaluation of artificial intelligence models for two-stage dental caries detection and classification. The images were retrospectively collected from the Department of Dental Radiology, Chevron Clinical Laboratory (Pte.) Ltd., Chattogram, Bangladesh, following institutional ethical approval. The dataset includes 3,002 annotated caries regions and 3,447 non-caries regions with bounding-box annotations in YOLO format. Before annotation, all radiographs were preprocessed using Contrast Limited Adaptive Histogram Equalization (CLAHE), bilateral filtering, and unsharp masking to improve image quality. Ground-truth annotations were independently created by three experienced dental professionals and finalized through expert consensus, achieving a Fleiss' kappa coefficient of 0.969. The dataset is intended for training, validation, and benchmarking deep learning models for automated dental caries detection and classification. It accompanies the research article: T.H. Shahim, Bhowmik S.R., Barua R., et al. "An Artificial Intelligence Approach Two-Stage Caries Detection and Classification using Panoramic Radiograph." Biomedical Signal Processing and Control (Under Review). Note: The complete dataset is not publicly accessible because of institutional and ethical restrictions related to patient privacy. Access to de-identified data may be considered upon reasonable request to the corresponding author and subject to approval from Chevron Clinical Laboratory (Pte.) Ltd.