Imaging performance of a LaBr<inf>3</inf>-based time-of-flight PET scanner

Abstract

There has recently been renewed interest in time-of-flight (TOF) PET due to the availability of fast scintillators that also have high light output and high stopping power, as well as cost-effective fast photomultiplier tubes and stable electronics. Early results with these TOF-PET systems have shown both an improved contrast/noise trade-off and faster convergence compared with reconstructions without TOF information. Simulations have predicted further improvement in imaging performance with better timing resolution. A prototype whole-body PET scanner incorporating Ce-doped LaBr <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</inf> crystals and specialized timing circuitry to take advantage of the scintillator’s fast timing characteristics has recently been completed. The intrinsic performance of the scanner has been measured. The average energy resolution over all crystals is 6.5%, and the system timing resolution is 375 ps. The scatter fraction for 20-, 27-, and 35-cm diameter cylinders is 21, 27, and 32%, respectively, for a 485-keV lower energy threshold. The average spatial resolution is 5.8 mm at 1 cm and 6.5 mm at 10 cm. Resolution modeling has been incorporated into the list-mode TOF iterative algorithm. Simulation studies were carried out to measure the relative impact of timing resolution, energy resolution and lower energy threshold, and spatial resolution modeling on TOF-PET imaging performance as characterized by the contrast/noise trade-off. It was found that improved timing resolution leads to faster, more uniform convergence with less variability in quantification as a function of either radial position or local activity environment. Better energy resolution allows for the use of a tighter energy window, which leads to fewer accepted scatter events and improved quantitative accuracy. Resolution modeling improves contrast recovery at the cost of slower convergence; further work is needed to define an accurate model of the point spread function for the LaBr <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</inf> system. The superior timing and energy resolutions appear to mitigate the loss of spatial resolution that arises from the lower stopping power of the crystal.