Breast Mass Segmentation in Mammographic Images Based on Gmac Model using Variational Level Set Method

Abstract

Breast cancer is the most common form of cancer in the female population. Mammography is recognized as the most effective method for its diagnosis. However, it is a very difficult, time-consuming to interpret mammographic images. Computer-aided detecting and diagnosis (CADx) systems have the potential to assist radiologists in the complex task of discriminating malignant and benign types of breast lesions and reducing the biopsy rate without increasing false negatives. In most types of CADx systems, a step is computing mammographic images features which automatically extracted from regions of interest (ROIs) depicting mammographic masses. Therefore, mass segmentation is a very important step for classification of suspicious regions as normal, benign or malignant, which influences the following classification and detection directly. Although so many mass segmentation algorithms have been proposed for mass segmentation in mammography, it is still a difficult and challenge problem to achieve accurate segmentation results. The purpose of this study was to develop an automated method for mammographic mass segmentation. The approach is based on Global minimum active contour (Gmac) model, and using variational level set method to solve the curve evolution problem derived by minimizing the curve energy. At last, a lot of experiments were carried out. Those segmentation results demonstrated the effectiveness of the proposed method.