Automated Classification of Breast Cancer Histology Images Using Deep Learning Based Convolutional Neural Networks

摘要

Automated classification of cancers using histopathological images is a challenging task of accurate detection of tumor sub-types. In this paper, we applied fine-tuned pre-trained deep neural networks classified on BreakHis datasets on eight distinct classes for benign has four sub-classes (adenosis, fibroadenoma, phyllodes tumor, and tubular adenoma) malignant has four sub-classes (ductal carcinoma, lobular carcinoma, mucinous carcinoma, and papillary carcinoma) all together on difference model on Inception (V1,V2) and ResNet V1 50. The confusion matrix showing high accuracy value 95% with less error rate 0.011 .