Recurrent Breast Cancer: Histopathological and Hyperspectral Images Database (HistologyHSI-BC-Recurrence)

This dataset contains 47 histopathological and 677 hyperspectral images from 47 patients diagnosed with breast tumors, together with their clinical data. Out of these patients, 22 had distant recurrence during the 10 years of follow-up after diagnosis, which is the Ground Truth for this dataset (recurrence vs no recurrence). The histopathological slides were stained with hematoxylin and eosin (H&E), digitized through the Pannoramic 250 Flash III scanner (3DHISTECH Ltd.) and annotated by skilled pathologists. Afterwards, these slides were captured at 20x magnification using a bright-field Olympus microscope equipped with a hyperspectral camera (Hyperspec VNIR) covering the spectral range from 400 to 1,000 nm. Distant recurrence is the leading cause of death in most cancers. Therefore, this database aims to be a tool to investigate the prediction of the distant recurrence, over 10 years of follow-up after diagnosis, from hyperspectral images of the primary breast tumor biopsies.