3D SURFACE RECONSTRUCTION OF LARGE MEDICAL DATA USING MARCHING CUBES IN VTK
W. Narkbuakaew, Saowapak Sotthivirat, Duangrat Gansawat, W. Areeprayolkij, Wasin Sinthupinyo, Sarin Watcharabutsarakham · 2008
A three-dimensional surface is a representation of volumetric image data in a shape form. One algorithm, which is acceptable for reconstructing a threedimensional surface, is the Marching cubes. It uses patterned cubes or isosurface to approximate contours. The marching cubes algorithm needs some processes or algorithms to reduce time and memory for reconstructing a surface from large volumetric data. A common way to solve this problem is by subsampling or reducing a volumetric image size, but the quality of the reconstructed three-dimensional surface will be poor if we only apply subsampling. Due to the effect of volumetric subsampling, we propose a process to improve the quality of a surface reconstructed from the sampled volumetric data. It is based on a pipeline of Visualization Toolkit (VTK). Our proposed approach includes three main steps, which are preprocessing, reconstructing, and displaying. In this paper, we focus on the preprocessing step including sampling, thresholding and Gaussian filtering. Specifically, we studied the effect of the subsampling factors, and the standard deviation parameters for Gaussian filtering. Furthermore, time and memory usages are discussed in this research as well. The experimental results show that the proposed process can be applied effectively to reconstruct a threedimensional surface from large volumetric image data.