Mitotic nuclei analysis in breast cancer histopathology images using deep ensemble classifier

Anabia Sohail, Asifullah Khan, Humaira Nisar, Sobia Tabassum, Aneela Zameer

Abstract: Highlights•A deep ensemble based automated system is developed for the analysis of mitoses in breast histopathology images.•In the first step, candidate mitoses from histopathology images are selected via deep instance segmentation model, whereas in the second step, deep CNNs based heterogeneous ensemble is developed for classification of mitotic nuclei.•Different architectural ideas in deep CNNs are exploited to capture the structural, textural, and morphological properties of the mitotic nuclei.•The proposed ensemble technique significantly improves the F-score with high precision and good detection rate.