Automated 3-D reconstruction of vascular structures from high definition casts

Publisher Website
Google Scholar
Abstract
Three-dimensional reconstruction and computer modeling is becoming recognized as a powerful tool for studying vascular structures. The computational approach, as well as the computer hardware selected for the task, however, depend upon the information desired. For the modeling of surface geometry, as in the case of the aortic valve, a surface formation technique is favorable over the more computationally demanding volume rendering approach. Automated surface formation, however, requires good quality, high contrast images. We therefore present a technique for producing high contrast images from high definition casts. We also describe the methodology used for automatic contour tracing, generating a mesh of variable density, and the schemes used to reconstruct bifurcating objects. With this approach, 98 mbytes of imaging data could be reduced to 180 kbytes of polygon vertices, and manipulated at near real-time speed on a medium performance graphics workstation. Such a system is therefore well suited for detailed, quantitative analyses of the reconstructed structures. Overall, this paper outlines the procedures used to create a high definition, three-dimensional computer model of any vascular structure.