Background: The Ordered Subsets Expectation Maximization (OSEM) reconstruction algorithm has been widely applied in emission CT reconstruction

such as SPECT

PET and synchrotron radiation X-ray fluorescence CT (XFCT). The quality of reconstructed image is better than other analytical methods such as filtered-back projection algorithm. However

the convergent rate of the OSEM is slow. Purpose: In order to speed up the convergent rate of OSEM

we want to investigate an improved OSEM algorithm. Methods: In this paper

we present an accelerated OSEM algorithm (AOSEM) by increasing the step size of the correction item and show its convergence characteristics with various subsets and accelerated factors. Results: The AOSEM algorithm is proposed and applied in XFCT image reconstruction. Comparing the two reconstruction algorithms

both the simulations and experimental results showed that AOSEM reached the same image quality as in OSEM but only about half the number of iterations when an accelerated factor was used. Conclusion: AOSEM can further speed up the convergence of OSEM when the accelerated factor h>1. Abstract