Deep learning-based computer-aided diagnosis (CAD) has been gaining popularity for analyzing histopathological images. However, there has been limited work that addresses the problem of accurately classifying breast biopsy tissue with hematoxylin and eosin stained images into different histological grades. We propose a system which can automatically classify breast cancer histology images into four classes, namely normal tissues, benign lesion, in situ carcinoma and invasive carcinoma. Our framework uses a Convolutional Neural Network (CNN) with a hierarchical loss, where failing to distinguish between carcinoma and non-carcinoma is penalized more than failing to distinguish between normal and benign or between in situ and invasive carcinoma. The network also includes a patch-wise design with global pooling directly on input images. By incorporating the hierarchical and global information of the input images, our framework can outperform the previous system by a large margin.
CITATION STYLE
Wang, Z., Dong, N., Dai, W., Rosario, S. D., & Xing, E. P. (2018). Classification of Breast Cancer Histopathological Images using Convolutional Neural Networks with Hierarchical Loss and Global Pooling. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 10882 LNCS, pp. 745–753). Springer Verlag. https://doi.org/10.1007/978-3-319-93000-8_84
Mendeley helps you to discover research relevant for your work.