Multi-Dimensional Transfer Functions for Tissue Selection in Computed Tomography

Computed Tomography (CT) is a widely used 3-D imaging technique. A 3-D volumetric grid is obtained from 2-D cross-sectional images. In order to be useful as a diagnostic tool, voxel-based numerical values that represent X-ray absorption in each voxel must be interpreted and combined to form an image that depicts tissue composition at a particular location. Transfer functions are used to translate measured X-ray absorption data into Hounsfield units, and Hounsfield units into intensities, colors and transparency values. Gradient-based transfer functions are used to highlight material boundaries and interfaces between different tissues. Multi-dimensional transfer functions combine the advantages of regular and gradientbased transfer functions, facilitating a wide spectrum of visual representations. Transfer functions are usually under user control and often difficult to find. Improper transfer functions can create misleading visualizations and may lead to erroneous diagnoses. This article discusses how multi-dimensional transfer functions can be derived that are clinically relevant and meaningful.