Datasets for Early Recurrence Prediction in Oral Squamous Cell Carcinoma

The oral tissue sections were stained with hematoxylin and eosin dye. We scanned the stained sections to obtain brightfield and confocal images. The images were annotated to acquire the region of interest (ROI). The ROI images were used to produce patches, and these were center-cropped and downsampled. All the image patches were saved in .png format.       To develop and validate deep learning models, the images were segregated at the patient level into 70% training, 10% validation, and 20% testing. The official code that uses this dataset is available on Multiple Instance Learning for Early Recurrence Prediction in Oral Squamous Cell Carcinoma.