Deep learning for the recognition of benign and malignant breast lesions in medical images

Abstract

The exploitation of medical imaging (“Big Data”) is currently a task that healthcare professionals find very difficult to manage. However, the creation of Intelligent Systems (IS) is nowadays an effective mean for the detection and the recognition of most frequent diseases, in particular breast cancer. In this study, we propose a new deep learning (DL) architecture for the recognition of malignant and benign tumors on breast biopsy images. The experimental results show that the proposed method (IDenseNet-BCBC for Improvement of DenseNet for Binary Classification of Breast Cancer) achieved an accuracy of 99.17% and 100% on magnified level 40x, and 100x, respectively.