Breast Density Analysis on Mammograms: application of Machine Learning with textural features

Breast cancer (BC) is the world’s most prevalent cancer in female population, with 2.3 million new cases diagnosed worldwide in 2020, but there has been an improvement in survival thanks to the greater diffusion of early detection programs and increasingly targeted treatments. Digital mammography (FFDM) is considered the cornerstone for diagnosis in the early stages of BC. From several previous studies it has emerged that breast density (BD) is a risk factor in the development of BC, affecting the periodicity of screening plans present today at an international level. In this study the focus is the development of mammographic image processing techniques that allow the extraction of indicators derived from textural patterns of the mammary parenchyma indicative of BD risk factor. Different Machine Learning (ML) techniques have been employed to classify breasts based on density. Textural features were extracted only from breast parenchyma with which to train classifiers thanks to aid of ML algorithms. The preliminary results are in line with those obtained in the literature, reaching an accuracy of 93.55% for a binary classification between dense and no-dense.