Computation of high overrelaxation parameters in iterative image reconstruction

Single-projection iterative reconstruction allows the use of high overrelaxation parameters which can be determined by postulating maximum increase of image quality during an iteration step. Further speeding is achieved if maximum increase of quality is postulated during two or more steps. This postulate leads to high initial overrelaxation values which drop down during the following steps. The use of parameters obtained by optimization during two steps requires about half the number of iteration steps compared to one-step parameters. Estimates are used to find the optimum parameters rapidly. Optimization within three or more steps provides only minor convergence improvement, but it demonstrates that different combinations of parameters may achieve images of similar quality. It follows that parameters derived from one data set may be widely used for other data. However, the optimizing procedure is expected to be helpful if data from new equipment or with new characteristics have to be reconstructed. Especially in three-dimensional iterative reconstruction, rapid convergence using high initial overrelaxation is important