ABSTRACT
Previous chapters have highlighted the two main categories of phantoms and the advantages and disadvantages of each. Brie y, existing computational phantoms involve a trade-off between realism and exibility, and this affects their applicability to imaging simulations. Since they are based on patient data, voxel-based phantoms are realistic, but they remain xed to a particular anatomy and resolution. Studies of the effects of anatomical variations or motion on medical imaging can be limited, and the generation of
the phantom at other resolutions requires interpolation, which introduces error. Stylized or mathematical phantoms, conversely, are de ned mathematically, in order to allow for anatomical variation and generation at multiple resolutions. The simplicity of the mathematical equations, however, limits an exact modeling of the organ shapes. In order to get a more realistic phantom, the sets of equations de ning the phantom have to become more complex; but with this increase in complexity comes a decrease in exibility. Current work in phantom development has focused on the development of “hybrid” phantoms that seek to combine the realism of a patient-based voxelized phantom with the exibility of a mathematical phantom. We have been leading the development of realistic and exible digital phantoms for use in medical imaging research. Figure 5.2 shows the evolution of computerized phantoms toward more ideal hybrid computational phantoms and summarizes our work in this evolution with the development of the mathematical cardiac-torso (MCAT), NURBS-based cardiac-torso (NCAT), extended cardiac-torso (XCAT), and mouse wholebody (MOBY) phantoms.
