VHVis was developed as a prototype using Java3D (SUN Microsystems), VTK (Kitware Inc.) and OpenGL APIs. A surface render algorithm and a volume render algorithm are currently supported. The downloaded VHIF-VOI files were used as input data. The symbolic anatomical hierarchy embedded in the files is automatically extracted and displayed. Each of these anatomical terms can be selected as an entity. Iso-surfaces of the anatomical structures are generated from the segmented images using marching cube algorithm. The surfaces can be smoothed using windowed-sinc filter. The extracted surfaces can be rendered as 3D-objects that can be interactively examined. Multiple 3D-anatomy structures obtained from different VHI-VOI files can be rendered and composed into a single scene according to their physical positions in the visible man. Multiple anatomical structures can be easily visualized using different material properties.

     The volume-rendering engine of VHVis was implemented using the 3D-texture-mapping algorithm. The RGBA 3D-texture map was built from the anatomical images and segmentation masks that contained in each of VHIF-VOI files. Opacity transfer function can be adjusted interactively using the segmentation mask as indices. VHVis also provides a hybrid model to mix the structures rendered using surface and volume rendering algorithms. Structures of interest are volume rendered and reference structures are quickly surface rendered. Cutting planes are supported to help view the inside structures.