Spiculated Lesions and Architectural Distortions Detection in Digital Breast Tomosynthesis Datasets

Digital Breast Tomosynthesis (DBT) is a new 3D imaging technique aiming at overcoming some limitations of mammography. A computer aided detection system may help the radiologist to process the increased amount of data of this new modality. In this paper we propose to address the detection of masses and architectural distortions in DBT datasets. To achieve this task, we propose a detection scheme composed of two separate channels, each of them being dedicated to the detection of one of the target radiological signs. We propose a description of these channels as well as a validation on clinical data. We also compare the performance with existing approaches.