Efficient Method for Generating a Pinhole Edge Penetration Model for Iterative Reconstruction using GATE

The ability to acquire images with resolutions in the micrometer range, as recently reported for pinhole and multipin-hole SPECT systems, raised interest in these imaging modalities within the preclinical small animal imaging research community. At such high resolutions, physical imaging properties such as pinhole edge penetration, detector blur, detector misalignment, the geometric response of the aperture, and gamma photon attenuation and scatter can severely degrade image quality. To improve system design and reconstruction methods, an aperture edge penetration model was developed. The model was simulated in GATE by replacing each imaging detector element with a point source and collecting the photons emitted by this source on the opposite side of the aperture. For efficiency, only one detector quadrant was evaluated and sampling of the model was performed for every fourth detector element. Furthermore, the photons were focused onto the aperture by adjusting the polar and azimuthal emission angles. Models were developed for two imaging systems using 2 different isotopes (Tc99m & I-125). The first system had a detector size of 260 mm and a pinhole-to-detector distance of 180 mm. The second had a detector size of 90 mm and a pinhole-to-detector distance of 100 mm. The same aperture was used for both systems. The system with the larger detector showed an enlargement in effective pinhole size inversely proportional to the distance between detector center and detector element, whereas the smaller detector showed a similar dependence at a reduced amplitude. For I-125, the effective pinhole size was similar to the bore diameter, since only a small number of the low energy photons penetrated through the aperture material. This simulation of the aperture edge penetration will aid in better understanding pinhole SPECT system design that minimize the penetration effects, as well as for development of more accurate reconstruction algorithms.