Mammograms are recommended for women beginning at age 40. New computer science methods are allowing us to decode the structure and texture of breast tissue seen on mammograms to better understand a woman’s risk of developing breast cancer. Our study looked at mammograms from women across multiple breast screening centers to explore whether patterns in the breast tissue—called parenchymal phenotypes—can give more information about breast cancer risk. Using advanced image analysis techniques (radiomics), we identified six consistent tissue patterns that were linked to future breast cancer diagnosis.
These tissue patterns were found to predict breast cancer risk in both Black and White women, independent of breast density. They also helped identify cancers that were missed during initial screening. By grouping texture features into meaningful patterns, we offer a new tool that could improve early detection and risk assessment—particularly in diverse populations.
This study was funded by an NIH grant.