A multi-scale approach to mass segmentation using active contour models

Abstract

As an important step of mass classification, mass segmentation plays an important role in computer-aided diagnosis (CAD). In this paper, we propose a novel scheme for breast mass segmentation in mammograms, which is based on level set method and multi-scale analysis. Mammogram is firstly decomposed by Gaussian pyramid into a sequence of images from fine to coarse, the C-V model is then applied at the coarse scale, and the obtained rough contour is used as the initial contour for segmentation at the fine scale. A local active contour (LAC) model based on image local information is utilized to refine the rough contour locally at the fine scale. In addition, the feature of area and gray level extracted from coarse segmentation is used to set the parameters of LAC model automatically to improve the adaptivity of our method. The results show the higher accuracy and robustness of the proposed multi-scale segmentation method than the conventional ones.