Investigation of time-of-flight benefits in an LYSO-based PET/CT scanner: A Monte Carlo study using GATE

Abstract

The advent of fast scintillators yielding great light yield and/or stopping power, along with advances in photomultiplier tubes and electronics, have rekindled interest in time-of-flight (TOF) PET. Because the potential performance improvements offered by TOF PET are substantial, efforts to improve PET timing should prove very fruitful. In this study, we performed Monte Carlo simulations to explore what gains in PET performance could be achieved if the coincidence resolving time (CRT) in the LYSO-based PET component of Discovery RX PET/CT scanner were improved. For this purpose, the GATE Monte Carlo package was utilized, providing the ability to model and characterize various physical phenomena in PET imaging. For the present investigation, count rate performance and signal to noise ratio (SNR) values in different activity concentrations were simulated for different coincidence timing windows of 4, 5.85, 6, 6.5, 8, 10 and 12 ns and with different CRTs of 100–900 ps FWHM involving 50 ps FWHM increments using the NEMA scatter phantom. Strong evidence supporting robustness of the simulations was found as observed in the good agreement between measured and simulated data for the cases of estimating axial sensitivity, axial and transaxial detection position, gamma non-collinearity angle distribution and positron annihilation distance. In the non-TOF context, the results show that the random event rate can be reduced by using narrower coincidence timing window widths, demonstrating considerable enhancements in the peak noise equivalent count rate (NECR) performance. The peak NECR had increased by ∼50% when utilizing the coincidence window width of 4 ns. At the same time, utilization of TOF information resulted in improved NECR and SNR with the dramatic reduction of random coincidences as a function of CRT. For example, with CRT of 500 ps FWHM, a factor of 2.3 reduction in random rates, factor of 1.5 increase in NECR and factor of 2.1 improvement in SNR is achievable. The results of this study show that in addition to the high sensitivity of Discovery RX PET/CT scanner, the implementation of TOF with proper CRT can efficiently improve the image quality in this scanner. Having successfully simulated the DRX scanner and utilization of TOF information, our research goal is to use the Monte Carlo simulation technique to arrive at powerful, accurate and feasible reconstruction algorithms.