Bilateral Asymmetry Detection in Mammograms Using Non-rigid Registraion and Pseudo-color Coding

Breast cancer is the most common cancer in women. The deviation from the normal architectural symmetry of the left and right breasts can indicate the abnormalities in early stage of breast cancer development. This paper presents a computer-aided detection algorithm for bilateral asymmetry in mammograms. First, the left and right mammograms are aligned by a novel non-rigid registration method which based on diffusion model with demons algorithm. It considers the moving image as a deformable grid which diffuses through the contours of the objects in itself. Then, after registration, a bilateral subtraction technique is used to enhance the conspicuity of possible masses. At last, a novel pseudo-color coding method proposed in this paper is used to further highlight the suspicious regions in the subtraction image. Results of the experiment based on mammograms from DDSM show that lesion areas on the processed pseudo-color images could be more easily distinguished than on the original gray ones. This preliminary study indicates that our methods is potentially useful as an aid to radiologists in the interpretation of mammograms.