Scarcity of publicly available oral cancer image datasets for machine learning research.

Publicly available image datasets of pathologies have easy accessibility and thus, are increasingly being used in the field of machine learning and medical diagnosis. As oral cancer is the most common head and neck cancer with an increasing incidence rate, it is of paramount importance to know the status of publicly available datasets. We designed a systematic search (PubMed, Google Scholar, Google Dataset Search, and Google) to identify the publicly available oral cancer image datasets. After carefully screening 332 articles/datasets, only one met the selection criteria and was available publicly. However, it contained images of cancerous lesions of only lips and tongue. This first-of-its-kind analysis made realize a dire need for publicly available datasets in oral cancer. It will help researchers in the development of effective machine learning algorithms for oral cancer.