A Segmentation Scheme of Brainstem and Cerebellum using Scale-Based Fuzzy Connectedness and Deformable Contour Model.

ABSTRACT

A semi-automatic multiple-stage segmentation scheme of brainstem and cerebellum is proposed in this paper. This proposed scheme contains a modified Scale-based Fuzzy Connectedness (FC) algorithm, a morphology operator, Chain code, and Active Contour Mode (ACM). A modified Scale-based FC is utilized to delivery a rough segmentation, a morphology operator and Chain code are used to compensate the uncompleted small regions and derive an initial contour, and ACM is employed to obtain the final object boundary with the aid of the initial contour from the previous stages. The experimental results verify that the segmentation results using the proposed scheme are more accurate than using a single algorithm based on a Scale-Based FC method.