Pectoral Muscle Detection in Digital Breast Tomosynthesis and Mammography

摘要: Screening and diagnosis of breast cancer with Digital Breast Tomosynthesis (DBT) Mammography are increasingly supported by algorithms for automatic post-processing. The pectoral muscle, which dorsally delineates the tissue towards chest wall, is an important anatomical structure navigation. Along nipple skin, muscle boundary often used reporting location lesions. It visible in mediolateral oblique (MLO) views where it well approximated a straight line. Here, we propose two machine learning-based to robustly detect MLO from DBT mammography. Embedded into Marginal Space Learning framework, involve evaluation multiple candidate boundaries hierarchical manner. To this end, novel method generation using Hough-based approach. Experiments were performed on set 100 volumes 95 mammograms different clinical cases. Our combined approach achieves competitive accuracy robustness. In particular, data, achieve significantly lower deviation angle error mean distance than standard proposed run within few seconds.