Towards a Computer Assisted Diagnosis System for Digital Breast Tomosynthesis

Recently, Digital Breast Tomosynthesis (DBT) has been established as a novel and promising imaging modality for the volumetric assessment of breast images, having the advantage, that mass lesions are not obscured by overlaying tissues. In order to support physicians during the assessment and diagnosis of breast lesions, Computer Assisted Diagnosis (CADx) Systems can be applied. In this work a prototypic CADx system for DBT data is described, consisting of several modules, including interactive lesion segmentation, volumetric lesion visualization, lesions characterization, and CBIR to search and retrieve most similar lesions from a reference image data collection of DBT datasets. Up to now, the inter- active volumetric lesion segmentation as well as the lesion characterization modules have been implemented and evalu- ated on real and synthetic DBT image data, respectively. It is expected that, if a high quality reference date collection of DBT data is available, that the results obtained in these initial experiments can be verified and confirmed.