Breast density is the major risk indicator for the breast cancer. The proposed work aims to develop a breast density classifier according to Breast Imaging And Reporting Data System (BIRADS) category based on the texture features on mammograms. The characteristics features of the image are defined to distinguish between breast density classes as dense and non dense. The features are extracted by the discrete orthonormal Stockwell transform (DOST) and local binary pattern (LBP). DOST feature can be highly effective discriminator for multi scale technique that extracts texture features of mammographic images. DOST is used for extracting macro features and LBP is used for extracting micro features. The results achieved can be proposed for breast density classification frame work that can be used in clinical environment for aiding the radiologists.
Volume 12 | 05-Special Issue
Pages: 1061-1066
DOI: 10.5373/JARDCS/V12SP5/20201857