Using greyscale voxel databases for improved shading and segmentation

Many different data representations are possible in computer graphics. Originally, in the medical field, simplified methods were used in order to reduce computation times on small computer systems. Currently a wider range of techniques is developing as costs of hardware continue to fall. In this paper we review a number of possible representations and explain the advantage of one that is greyscale, volumetric and random access. Different segmentation techniques can be used, as well as shading algorithms that give greatly improved appearances. A quantitative analysis of shading methods is derived in terms of the degree of sampling of the 'pseudo-normal' vectors that estimate the direction of the tangent to a surface. The application to a study of multiple sclerosis lesions in the brain using nuclear magnetic resonance data is shown.