Direct Volume Rendering of Photographic Volumes Using Multi-Dimensional Color-Based Transfer Functions

Traditionally, volume rendering of medical data has been comprised of transfer functions that map a scalar value, usually a MRI or CT intensity, to an opacity. Corresponding color maps are either selected regardless of the actual physical color of the volume (i.e. greyscale) or predetermined as in photographic volumes. Rarely has the voxel color been used as a means to define the associated opacity value. By using transfer functions that map multichannel color values(RGB or CIE L∗u∗v∗) to opacity, we can generate compelling and informative renderings that provide consistent boundary extraction throughout the volume. We present renderings of the Visible Human photographic volume using multi-dimensional color-based transfer functions. These functions were constructed by using gradient boundary enhancement techniques in conjunction with volume illustration techniques and incorporating the first and second directional derivatives along the gradient direction. We empirically compare the effectiveness of using the color-based transfer functions and discuss their applications and potential for future development.