A multi-focus oral panoramic x-ray image dataset based on pixel-level annotations

The dataset is a comprehensive, large-scale collection of panoramic X-ray images developed to advance research in dental artificial intelligence. It comprises 8,655 de-identified images and over 30,186 pixel-level annotations of lesion regions. These images were obtained from Changsha Stomatological Hospital with informed patient consent, and all personally identifiable information was removed. Annotation was performed manually by a team of 20 experienced dental imaging specialists using LabelMe software. The experts delineated individual tooth contours through polygonal annotations, targeting various common oral conditions. This rigorous process ensured high annotation precision and clinical reliability.The dataset supports multiple research applications, including tooth segmentation, lesion detection, and computer-aided diagnosis. Its effectiveness has been validated using several widely adopted deep learning models, demonstrating strong generalization capabilities and broad applicability.