Extraction of organs using region-growing method with knowledge of opposed domains and its application

摘要

The aim of this work is to extract organs in volume data with minimum numbers of user-defined models or parameters. In this report, we propose a region-growing method that utilizes the parameter calculated from the optimum path between a seed voxel in a target object and another voxel in opposed domains. The optimum path is defined using gradient between adjacent voxels along the path. We also apply the technique to volume visualization. Experimental results for an image of human embryo obtained with MR microscopy have displayed the effectiveness of this method.