Volume segmentation combined with morphology and Bayes
Zhi-Gang Sun, Jiawan Zhang, Jizhou Sun · 2004
A combination of morphology and Bayesian probability theory based on volume segmentation method is presented in this paper, which fully makes use of the volume histogram in the medical imaging data. There are two main steps in this segmentation framework. First, the morphological operations are applied by volume histogram to classify the different kinds of materials approximately which are used to calculate the classification parameters for the second step. Second, an accurate volume segmentation is generated by Bayesian probability theory. The voxel density distributions is modeled as a Gaussian distribution in which parameters is automatic and need no more training. The proposed segmentation method can not only efficiently implement in our medical application, but also automatically generate precise final results.