Methods for reconstruction of three-dimensional structures

The aim of the contribution is to provide the reader with an account of methods for creating computer reconstructions of medical data acquired in the form of planar cross sectional slices. This kind of data is frequently obtained from Various medical scanners like computed tomography and magnetic resonance imaging. The models of data are created by small volume elements in the shape of cube, called voxels. The voxels arranged in a regular space grid envoy information about the relevant volume of modeled object. Relating individual voxels to different parts of modeled modeled objects we define space structures (classification). The structures can be also chosen with thresholding. To visualize data we can chose two different approaches: volume or surface rendering. In the case of volume rendering the whole data set is visualized regardless to the classification and threshold. We described several methods: summed, averaged and maximum intensity projection. These methods are suitable for visualization of contrast structures. The second approach, surface methods, requires as the first step to define object - visualized structure either by thresholding (definition of voxels with intensity in given interval) or by the classification (definition of voxels with given volumetric properties). We described two simple algorithms (Voxel value shading, Depth shading) and several more complex shading ones (Z-buffer gradient shading, Voxel gradient shading and Gray level gradient shading). The setting of color and transparency to particular structures enables to visualize hidden or inner objects. The best quality of visualization is reached by Gray level gradient shading, when small surface details are Visible well.