Exploring the Best Parameters of Deep Learning for Breast Cancer Classification System

Breast cancer is one of the deadliest cancers in world. It essential to detect signs as early possible, make survival rate higher. However, detecting breast using machine or deep learning algorithms from diagnostic imaging results not trivial. Slight changes illumination scanned area can significantly affect automatic classification process. Hence, research aims propose an classifier for digital medical (e.g., Positron Emission Tomography PET, X-Ray Mammogram, and Magnetic Resonance Imaging (MRI) images). The proposes modified architecture with five different settings model classifiers. In addition, are also explored dataset used Curated Subset Digital Database Screening Mammography (CBIS-DDSM). A total 2,676 mammogram images split into 80%:20% (2,141:535) training testing datasets. demonstrate that trained eight layers Convolutional Neural Networks (CNN) (SET-8) achieves best accuracy score 94.89% 93.75% validation dataset, respectively.