Regularized Level Set Models Using Fuzzy Clustering for Medical Image Segmentation

Abstract

Level set methods are a kind of general numerical analysis tools that are specialized for describing and controlling implicit interface dynamically. It receives widespread attention in medical image computing and analysis. There have been a lot of level set models designed and regularized for medical image segmentation. For the sake of simplicity and clarity, we merely concentrate on our recent works of regularizing level set methods with fuzzy clustering in this paper. It covers two most famous level set models, namely Hamilton-Jacobi functional and Mumford-Shah functional, for variational segmentation and region competition respectively. The strategies of fuzzy regularization are elaborated in detail and their applications in medical image segmentation are demonstrated with examples.