Explainable classifier for improving the accountability in decision-making for colorectal cancer diagnosis from histopathological images

Patrik Sabol, Peter Sincák, Pitoyo Hartono, Pavel Kocan, Zuzana Benetinová, Alzbeta Blichárová, Ludmila Verbóová, Erika Stammová, Antónia Sabolová-Fabianová, Anna Jasková

Abstract: Highlights•An explainable classifier predicts eight tissue types of colorectal cancer images.•Classification performance enhanced using a fine-tuned Convolutional Neural Network.•Explanation interface for segmentation of the whole-slide histopathological images.•The system was designed to be used as a support system for medical experts.•14 pathologists consider the system useful and reliable.