SVD ANALYSIS OF BACKPROJECTING OPERATOR OF ML-EM PET IMAGE RECONSTRUCTION

  • Vencel Somai Budapest University of Technology and Economics
  • David Legrady Budapest University of Technology and Economics
  • Gabor Tolnai Budapest University of Technology and Economics

Abstract

We have demonstrated that a full Monte Carlo modelling in the backprojecting step of maximum likelihood expectation maximization image reconstruction of positron emission tomography may lead to deteriorated image quality by decelerating iteration convergence. The reasons behind this apparently anomalous behaviour is analysed in a simplified system by means of singular value decomposition and a new filtering technique is proposed for real systems. Some theoretical considerations for further development are also presented.

Author Biographies

Vencel Somai, Budapest University of Technology and Economics
Institute of Nuclear Techniques, researcher
David Legrady, Budapest University of Technology and Economics
Institute of Nuclear Techniques, associate professor
Gabor Tolnai, Budapest University of Technology and Economics
Institute of Nuclear Techniques, researcher
Published
2018-09-06
How to Cite
Somai, V., Legrady, D., & Tolnai, G. (2018). SVD ANALYSIS OF BACKPROJECTING OPERATOR OF ML-EM PET IMAGE RECONSTRUCTION. Radiology and Oncology, 52(3). Retrieved from https://radioloncol.com/index.php/ro/article/view/2891
Section
Medical Physics