Speaker
Dr
Pedro Guerra
Description
We have studied certain properties of the maximum likelihood expectation
maximization (ML-EM) algorithm for iterative image reconstruction in positron
emission tomography (PET). The principal aim of the work has been the development
and evaluation a new stopping criterion for this algorithm. We developed and used a
software platform based on Monte-Carlo techniques, which simulates the emission of
gamma rays in the source and their detection in the tomograph, to compute the
transition matrix and also to generate projection data for reconstruction. The
digital Hoffman brain phantom was used. A single-ring tomograph has been simulated
with 128 detector crystals on the ring, which has 15 cm radius and a field of view
(FOV) of 20x20 cm2. The images were reconstructed over a 64x64 and 128x128 grid. In
order to investigate the problem of the deterioration of the image quality after a
number of iterations, we have studied the statistical properties of the updating
coefficients in the ML-EM algorithm. The results of this study show that the values
of the updating coefficients for the non-zero reconstructed pixels follow a
distribution composed of a peak region around 1 which becomes progressively
narrower as the iteration proceeds, and also a tail below 1, which corresponds to
that part of the image that is far from being completely reconstructed. This tail
has an exponential form and is pushed to high values as the iterations progress. We
find that the minimum and average values of the tail are related to the quality of
the image. In particular for optimally reconstructed image these values are
independent of the number of counts and of the image size. The possibility of
exploring this property in stopping the algorithm at optimally reconstructed image
is discussed.
Author
Mr
Anastasios Gaitanis
(Department of Medical Physics, Medical School, University of Patras, Greece)
Co-authors
Dr
George Kontaxakis
(Universidad Politécnica de Madrid, E.T.S.I. Telecomunicación, Dpto. Ingeniería Electrónica, 28040 Madrid, Spain)
Prof.
George Panayiotakis
(Department of Medical Physics, Medical School, University of Patras, Greece)
Dr
George Spyrou
(Foundation for Biomedical Research of the Academy of Athens (IIBEAA), Soranou Efessiou 4, Athens 11527, Greece.)
Prof.
George Tzanakos
(University of Athens, Department of Physics, Division of Nuclear & Particle Physics, Panepistimioupoli, Zografou, Athens 15771, Greece.)
