Deep learning-based breast cancer detection using VGG-NiN architecture

Abstract

Breast cancer is a steadily growing disease that provides elevated mortality than other cancer types for global women. The survival rate of this cancer type can be uplifted through appropriate screening and diagnosis where the role of early detection of breast cancer is significant. In this regard, the design of Computer-Aided Diagnosis (CAD) with improved reliability is always in demand for earlier detection. The work proposed a VGG-NiN architecture integrating the pretrained CNN of VGG-16 together with spatial pyramid pooling (SPP) layer and network-in-network (NiN). The above-said layers are stacked to classify the breast cancer severities with minimum parameters and to speed up the model convergence and training process. For this, the work utilized the INbreast dataset for digital mammogram inputs. The obtained experimental results reveal the robustness of the proposed method as compared with the state-of-the-art CAD techniques.