Application of Fast-Marching method on mammograms for breast skin-line estimation
Roshan Yapa, Koichi Harada · 2007
Breast skin-line estimation and breast segmentation from its background is an important pre-process in mammogram image processing and computer-aided diagnosis of breast cancer. Ultimate goal of this pre-processing task is to recognize the breast tissue region on a mammogram and limit the area to be processed into a specific target region. This will increase the efficiency and accuracy of computer-aided diagnosis of breast cancer using mammograms. Also, the Fast-Marching is a method developed by J.A. Sethian. Fast-marching is a partial differential equation based numerical technique to track evolution of interfaces. This method has a higher potential of handling topological changes of evolving fronts naturally. In this paper we are discussing possibility of using fast-marching method for estimating breast skin-line and segmenting breast tissue region from its background. We have introduced some modifications to the traditional fast-marching method, specifically to improve the accuracy of skin-line estimation and breast tissue segmentation. In this experiment we have used 30 mammogram images from mini-MIAS mammogram image database to evaluate the results.