Cancer classification using deep learning from medical image data

Abstract

Cancer is a disease in which some of the body’s cells grow uncontrollably and spread to other parts of the body. Cancer can start almost anywhere in the human body, which is made up of trillions of cells. There is usually no cure for this disease and it is often believed to be untreatable. Breast cancer ranks second among the most fatal cancers, especially in women. Every year many women suffer and die because of breast cancer. Early detection of the disease can save many lives. Breast cancer screening with mammography is essential because it can detect any breast masses or calcifications early on. Because breast tissue is dense, detecting cancer mass is difficult, leading radiologists to use machine learning (ML) techniques and artificial neural networks (ANN) to speed up the detection of cancer. This paper explores the Mini DDSM dataset, containing 9698 digital mammogram images, which were augmented and preprocessed, and fed into CNN and MobileNet Architecture with the aim of detecting normal, benign and cancerous tissues with high accuracy. Therefore, our aim is to apply the deep neural network based algorithm on a cancer image dataset to classify cancer and take advantage of image analysis, pattern recognition, and classification processes, and then validating the image classification outcome against medical specialist expertise. The main objective of this research is to acquire a higher accurate outcome on detecting cancer from medical mammography. Index Terms— Breast cancer detection, neural network, Deep learning, Digital image processing.