On convergence analysis of an accelerated EM algorithm for SPECT imaging

ABSTRACT

The EM algorithm is an effective reconstruction method for SPECT, because it can incorporate various physical factors into reconstruction. However, the convergence rate is very slow. One of the accelerated means is increasing the iterative step length. Though it can speed up the iterative process, whether the modified algorithm will converge is still a question worth with in-depth exploration. In this article, we propose a modified accelerating algorithm and study its convergence characteristics in the modified iterative algorithm, and we also present a few numerical experiments.

Keywords:

• Convergence
• the EM algorithm
• KL-distance

2010 MR Subject Classifications:

• 44A15
• 65R10
• 68W25
• 65R32

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

The work was supported by the National Natural Science Foundation of China (No. 61271398) and K. C. Wong Magna Fund in Ningbo University.