Effect of geometrical constraints on PET performance in whole body simultaneous PET-MR

Simultaneous PET-MR scanners are being developed for whole body imaging. These systems require compact and MR compatible readout for the PET component. Another important modification is the geometry of the PET scanner which is determined by space constraints imposed by the surrounding MR scanner. The maximal radius of the PET scanner is limited and it becomes difficult to insert end shielding. The aim of this study is to determine the effect of modified geometry and reduced shielding on the PET performance with regards to spatial resolution, singles, trues, scatter and random coincidences. Materials and methods: All data were simulated using the GATE Monte Carlo simulation tool. The reference system for the simulation was a state of the art PET-CT scanner (Gemini TF scanner with LYSO crystals Philips Medical systems). This system has a diameter of 90 cm and end shields with an inner diameter of 70 cm. The energy resolution of the system is 12 % and based on this system a whole body PET scanner was designed with less modules positioned at a smaller radius. This modification enables it to fit inside a 3T MR scanner. This system was simulated without end shielding and with limited end shielding (60 cm diameter). For the three systems the trues, random and scatter were simulated to quantify the effect of the modified geometry. The object used was the 70 cm long NEMA scatter phantom containing activity in a line source at a radial distance of 4.5 cm. Results: Reducing the diameter from 90 cm to 70 cm results in an increase of the amount of trues by 28 %. The relative scatter fraction increases from 33 % to 36 % for the 70 cm diameter system without end shields. The introduction of short shields resulted in a small reduction (2 %) of scattered and random coincidence fraction. More detailed analysis about origin of the events showed that in the new design 85 % of scattered events originates from inside the FOV, while 90 % of the random coincidences is caused by outside FOV- - activity. Conclusions: For PET systems with good energy resolution, end shields only play a limited role in the reduction of scatter. The end shields are only blocking a limited part of the scattered outside FOV activity and are mostly effective in reducing the singles and resulting randoms from outside FOV.