Breast Density Classification Using Multifractal Spectrum with Histogram Analysis

This paper presents a novel method for breast density classification in digital mammograms using multifractal spectrum and histogram analysis. The multifractal spectrum is used to capture fibroglandular texture features in sub-images extracted from the breast region of mammograms, and feature vectors extracted from multifractal spectrum are used to classify sub-images into fat or dense category and to calculate breast percent density. For enhancing texture features and improving classification accuracy of breast density, standard deviation and skewness analysis of histogram on mammograms are considered. A full-field digital mammography (FFDM) dataset INbreast is used to test our proposed method and experimental results of breast density estimation outperform other reported methods, giving a higher classification accuracy (83.33%).