Breast Lesions Segmentation in Dynamic Contrast-Enhanced MR Images

Abstract

Breast cancer is a serious public health problem worldwide. An accurate segmentation of the breast lesion is important for clinical applications in monitoring tumor volume and in the quantification of tumor characteristics. Dynamic contrast-enhanced magnetic resonance imaging can be used in accurate segmentation and quantification of breast lesions. The aim of this review is to provide an overview of the methods of breast lesion segmentation methods in Dynamic Contrast-Enhanced MR Images. In this study, semi-automatic and automatic methods are reviewed. The algorithms that are employed in these articles are fuzzy c-means, neural networks, marker-controlled watershed, active contour model, Markov random field, connected component analysis, graph cut, region growing and level set. The fuzzy c-means clustering algorithm is a popular method used by most of the articles. A limitation of many studies is the small number of datasets. Although the datasets were not the same throughout the articles, the method used in Pang et al. [7] outperforms the other methods that showed both good and robust results and is potential for further research.