Monte Carlo simulations of the count rate performance of a clinical whole-body MR/PET scanner.

The combination of MR and PET scanners can provide a powerful tool for clinical diagnosis and investigation. Among the existing approaches, the most challenging is that of complete hardware integration of both scanners. Such an integrated tomograph would allow simultaneous acquisition of both modalities, which could help solve issues such as cardiac and respiratory motion. Full integration imposes restrictions on the design of the PET part, such as detector configuration and maximum ring diameter. Furthermore, MR components surrounding the PET detector ring may cause gamma ray interactions, thus affecting PET performance. The purpose of this article is to assess the performance of a hypothetical whole-body integrated MR/PET scanner using Monte Carlo simulation techniques and compare it to state-of-the-art PET/CT devices used in clinical routine. The Monte Carlo simulation toolkit used for this study is the GEANT4 application for emission tomography. A hypothetical whole-body MR/PET tomograph fully integrated at hardware level and positioned between gradient and local coils of the MR scanner has been modeled. The NEMA 2-2001 protocol has been used to configure the simulations in order to measure sensitivity, scatter fraction, count losses, and random detections. Global sensitivity values as a function of the lower-level discriminator (LLD) energy are provided for time resolutions of 5 and 2.25 ns. In addition, the scatter fraction of the system is studied as a function of the LLD for energy resolution values of 10%, 15%, and 20%. Finally, true, scatter, random, and noise equivalent count rate curves as a function of activity concentration are given for dead-time values of 136, 432, and 1150 ns and for time resolution values of 2.25 and 5 ns. The influence on the count rate performance of the integrated PET scanner of the new geometry and interfering MR elements has been measured. The results show that the interference of the MR components has a much lower impact than the reduction in the detector ring diameter. Due to the larger solid angle coverage, the sensitivity is higher than that measured for a clinical PET/CT system (6200-10 900 cps/MBq at the center of the scanner) but not enough to compensate the degradation of the noise equivalent count rate due to increased scatter detection. The simulations prove the viability of an integrated MR/PET system and suggest that priority has to be given to either the improvement of the temporal resolution or the correction of triple coincidences if competitive performance is to be achieved.