Abstract
In order to support doctors in medical care, a lot of systems and methods have been proposed. Our encompassing goal is to realize a neurosurgery simulation system, which is important in terms of encouraging young doctors. The system needs many kinds of geometric models, for example, a brain, a tumor and so on. The models should be constructed from patient's data for the accuracy of the simulation. In this paper, we propose 3D Active Sphere, which is a method to construct a 3D geometry model of a tumor automatically from a 3D image data (volume data) of patient's MR image data. Our method based on an energy-minimizing method is able to construct a geometry model by fitting vertices of the sphere model to a target region. By using 3D Active Sphere with the sphere model as similar to the region of a tumor, we are able to construct the model and reduce a calculation cost, but not by the previous method 3D Active Grid. We also try to construct the other shape model, which is not like a sphere, and show that the accuracy of a constructed model depends on the shape of the model used by the energy-minimizing method.
