Estimation of NECR, Scatter Fraction, and Sensitivity of a New MR Compatible Small Animal PET Insert Based on Monte-Carlo Simulations

We are currently designing an LYSO-coupled to SiPM PET insert for use in small animal PET/MR. The inner diameter of the insert is constrained to be no larger than 6.4 cm due to the allowable space in the 7T magnet that the insert is being design for. The design utilizes a dual layer offset crystal design to minimize the depth of interaction effect. Each detector block is simulated to be composed of a 10x10 / 9x9 array of 1.345x1.345 x 6/4 mm3 crystals (pitch = 1.422 mm) in the front/back layer. The tomograph is modeled as one, three, or six rings (ring pitch = 16 mm), leading to an axial extent of 14, 30, or 46 mm. GATE simulations were performed to estimate some performance characteristics of the tomograph NEC, sensitivity, and scatter fraction. Based on measured data from a single detector block, the energy resolution was estimated to be between 15 and 17%, the timing between 3 and 4 ns, and dead-time between 500 and 600 ns. A set of NEC curves was produced from the simulated data with different permutations of the above hardware characteristics. The simulations were made with a mouse-sized phantom as outlined in the NEMA NU4-2008 protocol. Simulations of a low activity point source in the centre of the tomograph were also made to estimate peak sensitivity. For a six-ring tomograph with the 250-750 keV window applied, it was found that peak NEC, peak NEC activity, and peak sensitivity would be 754 kcps at 56 MBq and 8.3% under the most ideal hardware conditions. With only three rings, these figures would be 245 kcps at 53 MBq, and 5.0%. When reduced to one ring, these figures would be 29 kcps, 54 MBq, and 1.9%. Regardless of the number of rings, scatter fraction was always similar to 16%. Results are also reported for the worst hardware configuration, and also with an energy window of 350 750 keV.