Accurate modeling of fan-beam collimators in brain SPECT imaging using Monte Carlo simulation

Iterative reconstruction algorithms allow compensation for the degradations suffered in the projection process. These algorithms require an accurate collimator point spread function (PSF) modeling which includes the detection system. The aim of this work is to model the fan-beam collimator PSF for a range of energies from 140 keV to 529 keV. Given a particular collimator configuration, a detailed hexagonal hole array was generated. Photon tracking inside the collimator was performed by using the Monte Carlo code PENELOPE. Four sets of counters were employed so as to differentiate contributions to the geometric, septal penetration, and coherent (Rayleigh) and incoherent (Compton) scatter components. Sensitivity and spatial resolution were calculated for each energy. Our results showed no significant differences in sensitivity (below 2%) for the source energies inside the detection window. Nevertheless, we observed significant differences when contamination of high-energy photons, like those of /sup 123/I, were considered. Our results suggest that PSF models studied with /sup 99m/Tc for fan-beam collimators cannot be used for brain SPECT imaging with other isotopes such as /sup 123/I.