Live-wire-based 3D segmentation method.

Abstract

Three-dimensional (3D) segmentation has become an essential part of volumetric image analysis. The methodology is based on the Live-Wire approach implemented in two-dimensional (2D) scene, yet two significant modifications have been employed. The wavelet-based cost map has improved the segmentation accuracy, whereas the Fuzzy C-Means (FCM) clustering procedure has shrunk the area to be searched, and thus has lower the numerical complexity. The 3D segmentation has been obtained by implementing the FCM clustering procedure with morphological operations. It does not require users interaction. The method has been employed to Magnetic Resonance Imaging (MRI) and X-Ray Computed Tomography (CT) studies. Additionally, noise resistance of the 3D approach has been measured.