Breast Cancer Detection Using Deep Learning

Abstract

Breast cancer, according to the Breast Cancer Institute, is one of the most difficult kinds of cancer that is extremely threatfulfor women all over the world. Early cancer discovery, followed by appropriate treatment, can help minimize the risk of deaths in the first stage of respective cancer only and increase the likelihood of survival by up to 8%. Medical imaging research has become increasingly reliant on machine learning and deep learning. In this research, we are utilizing a deep learning system to identify breast cancer on mammograms screening to get high accuracy. CNN and deep learning algorithms are commonly employed for image classification. The tumors are identified using a convolutional neural network approach. The final stage in determining whether the lesion under observation is normal or cancerous, is classification. If it is determined to be a cancerous component, additional categorization is performed to establish cancer's future behavior, i.e., benign or malignant. Steps like segmentation and feature extraction are required for classification. Our study made use of mammograms from the Mammographic Image Analysis Society (MIAS) dataset that contains a total of 322 mammographic images. It is a work in progress, and additional progress is being made by optimizing the CNN architecture and using pre-trained networks, both of which should result in greater accuracy metrics.