Abstract
The heuristic reconstruction process used for the highly focused, scanned focal point tomographs from 1978 and described by the authors in 1979 was changed to a much more accurate two-dimensional iterative reconstruction in 1988. The new algorithm is specific to the unique geometry of the multidetector tomograph and produces dramatically improved images that approximate least mean-square estimates of the objects. This paper describes the algorithm and its generalization to three dimensions. The handling of attenuation, scattering and object power density spectra are discussed. Results are shown as a sampling of 3D brain phantom and clinical images.