Influence of detector thickness on resolution in three-headed gamma camera PET

In PET imaging with large area NaI scintillators, uncertainty in the measured position of a photon hitting the detector at an oblique angle exists due to the thickness of the crystal in the detector. To evaluate the influence of the crystal thickness and the angle of incidence on the spatial resolution, simulations, and measurements of a dual- and triple-headed PET system in different configurations were performed. In the simulations, the resolution was calculated as a function of: head radii, crystal thickness, head orientation, and the maximum angle of incidence. The influence of these parameters (except crystal thickness) was also evaluated by measurements on a dual-headed AXIS and on a triple-headed IRIX PET system. It was observed that reducing the average angle-of-incidence (and, hence, reducing the depth-of-interaction uncertainty) tended to improve the spatial resolution. This could be accomplished by restricting the maximum angle-of-incidence, increasing the separation of the heads or changing the orientation of the heads (a 0-120-240 configuration has a greater average angle-of-incidence than a 0-90-180 or 0-180 configuration). However, when taking the effect of noncolinearity of the annihilation into account, increasing the separation of the heads results in a degradation of resolution. The resolution improved by decreasing the crystal thickness (since depth-of-interaction uncertainty was reduced). The resolution was worst for all parameters expected to yield the highest sensitivity (a thick crystal, small radius, and a large acceptance angle).