Surface shading in tomographic volume visualization: a comparative study

The quality of different shading algorithms for use in tomographic volume visualization (z-buffer gradient, gray-level gradient, adaptive gray-level gradient, marching cubes, marching cubes segmentation with Phong interpolation, transparent gray-level gradient), using simulated and clinical data, is investigated. Error images are used to visualize deviations of the calculated surface normal vectors from the ideal values. For the assessment of transparent rendering, reflection maps are used. The best results are obtained with adaptive gray-level gradient shading and marching cubes segmentation with Phong interpolation. The first method is based on the voxel model and thus is especially suitable for interactive data exploration. The second creates a polygon structure, which may be manipulated with methods developed in computer graphics. Transparent gray-level gradient shading should only be used if no binary segmentation can be done