DETECT97 simulation studies of light output in a full field-of-view small gamma camera

Motivated by the objective of designing a full field-of-view (FOV) small gamma camera for medical applications, the authors investigated the limit on spatial resolution imposed by the gamma radiation detection components of 3 variations of one such device. For their investigation, the authors used a parallelized version of DETECT97, a Monte Carlo optical photon tracking simulator. The basic component of the device simulated was an assembly comprising a 100 mm/spl times/100 mm/spl times/8 mm NaI(Tl) scintillation crystal coupled to a block of quartz. The authors generated the point spread functions of 3 different crystal/quartz assemblies. The physical parameters that differed among the 3 assemblies were the quartz thickness and the crystal/quartz surface reflection coefficient. The authors performed simulations with scintillation photon generation initiated at various lateral positions within the crystal of each assembly. They emphasized positions at the periphery of the FOV of each assembly. To determine the best spatial resolution achievable by an unbiased estimator, the authors computed the Cramer-Rao lower bound on the variance of the position estimate for the simulated models and positions. Their computations indicate that at the centre of the FOV of each assembly the lower bound on spatial resolution is less than a millimeter and that the lower bound on resolution is less than 4 mm to within 7 mm of a corner of a crystal if light reflection from the side of the crystal is minimal. If light reflection from the side of a crystal is substantial, the lower bound on resolution is similar to the above in the centre of the crystal but degrades substantially 10 mm from the corner of the crystal. The authors' computations also indicate that the estimation error correlation coefficients are generally high (|corr|-1) for peripheral source positions.