BACH: Grand Challenge on Breast Cancer Histology Images.

•The BACH challenge was organized to push forward methods for automatic classification of breast cancer biopsies using clinical hematoxylin-eosin stained histopathological images.•A large public dataset, composed of 400 microscopy images and 30 whole-slide images, was specifically compiled for the BACH challenge.•A total of 64 methods were submitted, out of 677 registration, and a detailed comparative analysis was carried out for the methods with higher accuracy scores.•Several submitted algorithms performed better than the state-of-the-art in terms of accuracy (top score of 87%).•Convolutional neural networks dominated the submissions, and was the method of choice in the algorithm that won the challenge.