ABSTRACT
This study investigates the physical limitations involved in the extraction of accurate timing information from pixellated scintillation detectors for positron emission tomography (PET). Accurate physical modeling of the scintillation detection process, from scintillation light generation through detection, is devised and performed for varying detector attributes, such as the crystal element length, light yield, decay time and surface treatment. The dependence of light output and time resolution on these attributes, as well as on the photon interaction depth (DoI) of the annihilation quanta within the crystal volume, is studied and compared with experimental results. A theoretical background which highlights the importance of different time blurring factors for instantaneous ('ideal') and exponential ('realistic') scintillation decay is developed and compared with simulated data. For the case of a realistic scintillator, our experimental and simulation findings suggest that dependence of detector performance on DoI is more evident for crystal elements with rough ('as cut') compared to polished surfaces (maximum observed difference of 64% (25%) and 22% (19%) in simulation (measurement) for light output and time resolution, respectively). Furthermore we observe distinct trends of the detector performance dependence on detector element length and surface treatment. For short crystals (3 × 3 × 5 mm(3)) an improvement in light output and time resolution for 'as cut' compared to polished crystals is observed (3% (7%) and 9% (9%) for simulation (measurement), respectively). The trend is reversed for longer crystals (3 × 3 × 20 mm(3)) and an improvement in light output and time uncertainty for polished compared to 'as cut' crystals is observed (36% (6%) and 40% (20%) for simulation (measurement), respectively). The results of this study are used to guide the design of PET detectors with combined time of flight (ToF) and DoI features.
