| 摘要 |
<P>PROBLEM TO BE SOLVED: To provide a method to enable consistent segmentation of target objects. <P>SOLUTION: The method comprises obtaining data from a computerized tomography apparatus, receiving voxel cluster selected by an user with a target object included in an initial image, deriving the initial segmentation of the target object, herein this initial segmentation being obtained based on remarkable luminance of the target object, drawing the initial image based on the initial segmentation, deciding the main mode from the prior segmentation through a mean shift algorithm including the evaluation of voxel luminance transition decided in the above initial segmentation, providing the main mode to the algorithm adapted to derive the improved image of the target object, and drawing the improved image of the target object. <P>COPYRIGHT: (C)2007,JPO&INPIT |
