A Monte Carlo Simulation Study of the Optimal Source Offset for Scatter Fraction Estimation in PET and the Influence of the Scanner and Object Characteristics

Abstract

In positron emission tomography (PET), the scattered photons represent a major image degrading factor. The scatter fraction (SF) of PET scanners is experimentally determined following the National Electrical Manufacturers Association (NEMA) NU 2 protocol, using a polyethylene phantom with a line source inserted at 45 mm radial offset from the phantom’s center. In this work, we aim to investigate the optimal radial offset of the line source, that represents the SF of a uniformly activated phantom, for the lutetium yttrium orthosilicate (LYSO)-based uEXPLORER total body PET scanner. Additionally, we propose to study the impact of several factors (axial field-of-view (AFOV), crystal material and heterogeneities) on this optimal offset, in an effort to provide recommendations for a wide range of PET scanner configurations. We performed Monte Carlo (MC) simulations of the uEXPLORER scanner along with five phantoms of different diameters, using the Geant4 Application for Tomographic Emission (GATE) code. We performed supplemental MC simulations in which we varied the crystal material, the AFOV and we introduced bone and lung heterogeneities in the phantoms to assess their impact on the optimal offset findings. Our results demonstrated that the 45 mm offset overestimates the SF of the uniform 20 cm diameter phantom by 24.3\(\%\). Instead, an offset of 87 mm was reproducing this SF. For the phantoms having diameters in the range of 10–50 cm, an offset of 90.7\(\%\) of the phantom’s radius was shown to be optimal (best fit). No AFOV-related impact has been observed on this optimal offset. The influence of the heterogeneities was also small with less than 1.1\(\%\) absolute deviation in the SF. This offset could then be used for all AFOV lengths PET scanners using LYSO crystal. The impact of the crystal material was shown to be minimal and an offset of 91.5\(\%\) of the phantom’s radius is consequently recommended as an average for other PET scanners.