3D level set model for medical image segmentation
Guisheng Yin, Ying Lin, Yuhua Wang · 2009
Three-dimensional segmentation of medical volumetric image data as a basis of 3D reconstruction has important significance in biomedicine engineering. However, noises or intensity inhomogeneity in practical application often make 3D medical images segmentation become formidable. To effectively alleviate these problems, this paper presents a novel variational level set framework using neighbors statistical analysis. Firstly, a basic 3D level set model is constructed based on Bayesian inference for the segmentation of objects from 3D volumetric image data. Then neighbors statistical analysis is introduced into above model in order to overcome disturbances caused by noise and intensity inhomogeneity. Experiments have demonstrated that the proposed method performs well in 3D volumetric data segmentation in intensity inhomogeneity and noises scene.