Novel Local-Region-Based Active Contour Integrating Fuzzy Clustering and Structure Constraint for Putamen Segmentation in T1-Weighted Magnetic Resonance Brain Images

摘要

In this paper, we propose a novel putamen segmentation method in T1-weighted magnetic resonance (MR) images based on the local-region-based active contour model (ACM). Most existing image segmentation methods do not consider the complex background around the putamen during putamen segmentation, which may not be robust enough for segmenting the putamen in MR brain images. To improve putamen segmentation, we integrate the fuzzy cluster scheme and the structure constraint term into local-region-based ACMs. In particular, the fuzzy cluster scheme is introduced into the computation of the local version of intensity, which has the ability to solve the putamen segmentation in the presence of complex background that insula and globus pallidus cause. Moreover, considering the connection between putamen and claustrum in MR brain images, the structure constraint term is proposed to enhance putamen segmentation because only the local intensity information is not qualified for putamen segmentation in this case. Our method has been validated for segmenting putamens in T1-weighted MR brain images, and the experimental results have demonstrated that our method could achieve the promising segmentation performance.