ORALIS-Oral Region Annotated Learning Image Set for AI-based Oral Health Research

ORALIS: Oral Region Annotated Learning Image Set for AI-based Oral Health Research Description: ORALIS is a comprehensive and annotated intraoral image dataset developed to advance AI-based oral health research. It includes expert-labeled images across FOUR diagnostic categories—normal, variation from normal, OPMD, Oral cancer—covering a range of clinically relevant oral conditions. The dataset is intended to facilitate the training and evaluation of deep learning models for image classification, segmentation, and automated diagnosis. ORALIS provides a valuable benchmark for researchers, educators, and AI developers working on innovative solutions in dentistry and public health. Data collection: Data collection was done at one of the spokes - Priyadharshini Dental college and Hospital, Thiruvallur,Chennai engaged by Ragas dental college and hospital,Chennai-(HUB) employing hub and spoke model under Indian Council of Medical Research (ICMR) funded project. The copyrighted standard operating procedures (SOP) for taking smartphone-based intraoral photographs was used in capturing intraoral images. The SOP was curated by experts in the field to ensure the reliability, reproducibility and validity of dataset images. Ethical approval for data collection was secured in advance from the Institutional Ethical Review Committee (Approval No: RIEC/20231021/PHD). Additionally, informed consent was obtained from all participants, and data protection and patient confidentiality were maintained in accordance with the Helsinki Declaration. Identifying details of the patients have been removed from the metadata to protect the privacy and confidentiality of the patients. Image Annotations: This dataset comprises 190 patients and includes a total of 1520 images. To ensure thorough coverage, dorsal and ventral surfaces of the tongue, the left and right buccal mucosa, as well as the upper and lower dental arches were captured using smartphone. Each image was renamed based on the patient ID, and annotations were performed using the VGG Image Annotator tool (version 3.0.13). Skilled dentists used the VIA tool's features to annotate all images, identifying various lesions and oral structures. To enhance the dataset's comprehensiveness, each patient's data was integrated with annotated images, unannotated images, working JSON file and patient metadata in an Excel sheet. Funding This intraoral image dataset is a part of a study funded by the Indian Council of Medical Research (ICMR), New Delhi, under the Investigator-Initiated Research Proposals – Small Extramural Grants, 2023 (Project ID: IIRP-2023-1049).