SELECTION OF TASK-DEPENDENT DIFFUSION FILTERS FOR THE POST-PROCESSINGOF SPECT IMAGES

Iterative reconstruction from single photon emission computed tomograp hy (SPECT) data requires regularization to avoid noise amplification a nd edge artefacts in the reconstructed image. This is often accomplish ed by stopping the iteration process at a relatively low number of ite rations or by post-filtering the reconstructed image. The aim of this paper is to develop a method to automatically select an optimal combin ation of stopping iteration number and filters for a particular imagin g situation. To this end different error measures between the distribu tion of a phantom and a corresponding filtered SPECT image are minimiz ed for different iteration numbers. As a study example, simulated data representing a brain study are used. For postreconstruction filtering , the performance of 3D linear diffusion (Gaussian filtering) and edge preserving 3D nonlinear diffusion (Catte scheme) is investigated. For reconstruction methods which model the image formation process accura tely, error measures between the phantom and the filtered reconstructi on are significantly reduced by performing a high number of iterations followed by optimal filtering compared with stopping the iterative pr ocess early. Furthermore, this error reduction can be obtained over a wide range of iteration numbers. Only a negligibly small additional re duction of the errors is obtained by including spatial variance in the filter kernel. Compared with Gaussian filtering, Catte diffusion can further reduce the error in some cases. For the examples considered, u sing accurate image formation models during iterative reconstruction i s far more important than the choice of the filter.