System modeling studies in iterative X-ray CT reconstruction

Iterative reconstruction shows potential for noise reduction and resolution improvements in X-ray CT. The forward projection operator is a key modeling choice in every iterative reconstruction method. The operator may be expressed in terms of the basis functions corresponding to a continuous representation of the image space. In this paper, we study three system models describing the mapping from image to sinogram space: the square pixel/cubic voxel basis, the Kaiser-Bessel ldquoblobrdquo function proposed by Lewitt, and DeMan & Basu´s computationally efficient distance-driven method. We compare their behavior in reconstruction of synthetic phantoms and helical scan data in two and three dimensions.