Anisotropic diffusion tensor applied to temporal mammograms: an application to breast cancer risk assessment

Breast density is considered a structural property of a mammogram that can change in various ways explaining different effects of medicinal treatments. The aim of the present work is to provide a framework for obtaining more accurate and sensitive measurements of breast density changes related to specific effects like Hormonal Replacement Therapy (HRT) and aging. Given effect-grouped patient data, we demonstrated how the diffusion tensor and its coherence features computed in an anatomically oriented breast coordinate system followed by statistical learning scheme provides non subjective and reproducible measure, as compared to the traditional BIRADS and computer aided percent density measure. We also demonstrate how orientation of breast tissue changes in temporal study. This framework facilitates radiologist to assess breast tissue change and guide them to evaluate individual risk of having breast cancer.