Analysis of The Various Techniques Used for Breast Segmentation from Mammograms

Abstract

Studies show that the cancer that causes the breast is the most frequent type of cancer found among women. X-ray imaging, called mammography, is an imaging technique that is commonly used to detect and classify breast abnormalities. However, accurate segmentation of breast tissues and abnormalities in the mammogram is a challenge, and consequently, many techniques have been employed over the years to extract these tissues and abnormalities and classify breasts based on their vulnerability to breast cancer. In this paper, we present different approaches used for breast segmentation from mammograms. Various methods ranging from modern deep learning-based techniques like UNet, and Atlas-based techniques are reviewed, and the classical techniques such as active contour, global threshold, machine learning based methods, etc. The results of these techniques are compared in order to provide an insight into the challenges of breast tissue classification and the future challenges are highlighted.