MeMoSA dataset: A multi-country collection of over 30,000 oral mucosa images with clinically labelled lesions

Résumé fourni par la source

The rising incidence of oral cancer and associated poor prognosis, primarily due to delayed diagnosis, highlight the urgent need for artificial intelligence tools in clinical detection. However, efforts in this regard are hampered by the lack of large and ethnically heterogenous image datasets of oral lesions with clinically validated diagnoses. To address this gap, oral mucosa images captured with mobile device cameras were collected from cohorts spanning five countries. The images were systematically annotated with lesion type classifications as well as specific clinical diagnoses, then assessed for quality. The diagnoses were verified retrospectively by biopsy, where applicable, or by consensus verification by dental experts. The final dataset consists of 30,039 oral mucosa images supplemented by clinical metadata, made available on the MeMoSA Workbench platform. We believe that the MeMoSA dataset will serve as a significant resource to drive the training, evaluation, and refinement of AI-driven diagnostic algorithms, potentially improving diagnostic accuracy and enabling rigorous benchmarking against clinical expert assessments, for the early detection of oral cancer.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.