engleski

Deformable models for medical image segmentation

In recent years, significant research efforts have been dedicated to deformable model-based image analysis. In this work a method for image segmentation of abdominal aortic aneurysm is presented. Evaluation of abdominal aortic aneurysm is important for stent placement. In this paper we present a method for segmentation of abdominal aortic aneurysm based on a deformable model. In this approach, a deformable closed structure is placed on the image and is deformed according to equations of non-rigid body motion in a gravitational field, where gradient or some similar feature of the image represents height. After deformation the model is shaped as the image region (organ or tissue) which has to be segmented. In implementation of the deformable model we use the level set approach which has several advantages over the standard snake implementation. We apply this method to segment 2-D CT medical images of aortic aneurysm with encouraging results. A 3-D approach where deformable structure is a 3-D object (or balloon) is currently under development. 3-D deformable models have several advantages over 2-D models, because they can utilize additional 3-D information about the object. The latest results of aneurysm segmentation in 2-D and in 3-D will be presented in the paper.

active contour; image segmentation; aortic aneurysm

nije evidentirano