ABSTRACT
Cho et al have proposed a fast backprojection scheme for parallel beam geometries, the incremental algorithm, which performs backprojection on a ray-by-ray (beam-by-beam) basis as opposed to a pixel-by-pixel approach. We present an extension of this incremental, beamwise, backprojection algorithm to the case of volume reconstruction of data acquired by a cylindrical, multiring positron tomograph. Use is made of geometrical symmetries of the image volume. This method results in a twelve-fold reduction of execution time compared with a straightforward, voxel-driven implementation of the same interpolation equations.
Subject(s)
Tomography, Emission-Computed/methods , Algorithms , Diagnostic Imaging/methodsABSTRACT
Various analytical, iterative and rebinning algorithms have been proposed for 3D reconstruction in positron emission tomography. This paper examines execution times of five analytical and rebinning algorithms. Meaningful comparisons are obtained by using similar software modules in all implementations. Reconstruction times are shown to differ by vast amounts: the Favor algorithm of Defrise et al is more than twice as fast as the widely used reprojection algorithm of Kinahan and Rogers; the Fourier rebinning algorithm (Fore) recently developed by Defrise is more than 15 times faster than the reprojection algorithm.
Subject(s)
Algorithms , Image Processing, Computer-Assisted/methods , Image Processing, Computer-Assisted/statistics & numerical data , Tomography, Emission-Computed/statistics & numerical data , Biophysical Phenomena , Biophysics , Humans , Time FactorsABSTRACT
O problema de tempos longos para reconstrução em PET foi abordado de várias formas, resultando em um sistema dedicado, capaz de efetuar reconstruções 3D rotineiras em poucos minutos por frame: a abordagem por hardware usa processadores rápidos e arquitetura paralela; a abordagem software utiliza a implementação eficiente de algoritmos mais rápidos.
The issue of long reconstruction times in PET has been addressed from severa! points of view, resulting in an affordable dedicated system capable ofhandling routine 30 reconstructions in a few minutes per frame: on the hardware side using fast processors and a parallel architecture, and on the software side, using efficient implementation of computationally less intensive algorithms.
