NLM Home Page VHP Home Page

Next: Full Text Index Contents: Conference Page 

Volume Segmentation of the Visible Human CT Data Set

Wenjian Wang, William G. Wee, Xun Wang
Artificial Intelligence and Computer Vision Laboratory
Department of Electrical and Computer Engineering and Computer Science
University of Cincinnati, Cincinnati, OH45221
wwang@ececs.uc.edu, wwee@ececs.uc.edu, xwang@ececs.uc.edu
Abstract
      This research is a feasibility study of developing a semi-automatic (with a minimum human intervention) data segmentation system to facilitate segmentation of anatomical entities in visible human data set, quantitative analysis of human body parts, and visualization of human anatomy. The ultimate objective is to extend the above results to living human data and to facilitate a feasibility study of virtual surgery. The input data to our system are CT images of the visible human fresh cadavers. A thresholding operation is first employed to remove background noise from these images. Then two methods are employed to extract 3-D surfaces: a facet model method, and an active contour model method. The facet model method employs a 3-D facet model to extract 3-D surface points directly from a series of CT images with subpixel accuracy. The active contour model method first extracts edge points of an interested anatomical entity from each slice of image and then stacks these edge points together to form 3-D surfaces. We have successfully extracted the external surface and the skull from the CT images using the above methods. All extracted entities are volume-displayed.
 
Keywords: data segmetation, 3D surface, contour detection, subpixel accuracy.

Table of Contents

Next: Full Text Index Contents: Conference Page