A dynamic interactive segmentation for medical images

A medical image is one of most complicated images and its segmentation is crucial for sequential analysis and diagnosis. In this paper, a new method based on maximum-gradient and orientation-code is presented. When the approximate positions of a few points along the border are located, it adopts a maximum-gradient and orientation-code as heuristic knowledge to search for the optimal border. Moreover, some rules are defined. If some segments of the border is thought to be divergent, it can adjust it by dynamically modifying some parameters according to those rules. Therefore, the segmentation procedure is dynamic and interactive. This method is shown to be efficient and accurate in cardiac imaging.