Skip to main content
. Author manuscript; available in PMC: 2014 Feb 1.
Published in final edited form as: IEEE Trans Nucl Sci. 2013 Feb;60(1):166–173. doi: 10.1109/tns.2012.2233754

Fig. 4.

Fig. 4

Time performance for the FFT and the CUDA approach. [top] time performances for different SR kernel resolutions and for a fixed voxel size and dimension (4 mm3, 144 x 144 x 48); [top-left] varying TOF resolution (300 to 1200 ps FWHM) with fixed LOR resolution (10 mm and 5 mm radial and axial FWHM, respectively); [top-right] varying LOR resolution in the radial direction (5 to 20 mm FWHM) with fixed TOF (300 ps) and axial (5 mm) resolutions. [bottom] time performances for different voxel sizes (4 to 2 mm), and fixed SR kernel resolutions (375 ps FWHM for TOF and 6.5 mm for LOR) and voxel dimension (144 x 144 x 48). The red rectangle indicates configurations with parameter settings that are clinically practical.