Adaptive Region Growing Algorithm in Medical Images Segmentation

In this paper we put forward an adaptive region growing algorithm for image segmentation by estimating the parameters through investigation of the statistical characteristics in local regions. By the algorithm, we use Gaussian mixture model to describe the properties of the region and employ an effective data clustering algorithm to calculate the parameters. Considering the similarity of the neighbor slices in the medical volume data, a Gaussian filter is applied in perpendicular direction to reduce the noise. In a slice image, an anisotropic diffusion filter is used to preserve the edge information. A number of medical images are tested to demonstrate the applicability and reliability of the proposed algorithm.