The one-block version of ordered subsets (OS) techniques was used to accelerate the convergent rate of the space-alternating generalized expectation-maximization (SAGE) algorithm. The new row-action SAGE (RA-SAGE) algorithm processed projections in sequentially orthogonal order which reduced the dependency among the projections and speeds up the convergences. Additionally, the over-relaxation parameter in the direction defined by the RA-SAGE algorithm was also applied to obtain fast convergence to a globally maximum likelihood (ML) solution. In experiments, the RA-SAGE algorithm and the classical SAGE algorithm were compared in the application to positron emission tomography (PET) image reconstruction. Simulation results showed that RA-SAGE had better performance than SAGE in both convergence and image quality.
A new method to accelerate the convergent rate of the space-alternatinggeneralized expectation-maximization (SAGE) algorithm is proposed. The new rescaled block-iterativeSAGE (RBI-SAGE) algorithm combines the RBI algorithm with the SAGE algorithm for PET imagereconstruction. In the new approach, the projection data is partitioned into disjoint blocks; eachiteration step involves only one of these blocks. SAGE updates the parameters sequentially in eachblock. In experiments, the RBI-SAGE algorithm and classical SAGE algorithm are compared in theapplication on positron emission tomography (PET) image reconstruction. Simulation results show thatRBI-SAGE has better performance than SAGE in both convergence and image quality.
