A robust breast segmentation method to support computer aided evaluation and breast density assessment

Introduction Advanced computer assisted image evaluation of breast cancer requires separation of breast tissues from other tissues and regions of the body, such as chest muscle, lungs, heart and ribs, that may confound image analysis. In addition, MR breast images generally have inhomogeneous signal intensities, together with partial volume effects at tissue or skin interfaces, that can compromise the performance of automated image processing methods used for tissue classification, patient motion correction or lesion localization. In this study, we aimed to develop a robust semi-automated algorithm for the segmentation of breast tissues.