Monte Carlo Modeling of Cascade Gamma Rays in PET

We have developed a new Monte Carlo simulation based on SimSET to model spurious activity caused by cascade gamma (gamma) rays in PET imaging. A torso phantom including lungs, a spine and a lesion was simulated to mimic PET imaging of Y-86. Contributions from positrons (beta(+)) only, beta(+)-gamma and gamma-gamma cascade events were binned in separate 2D sinograms and analyzed in the reconstructed images. The total abundances for beta(+) only, beta(+)-gamma and gamma-gamma cascades events modeled are 2.5%, 30.5% and 60.9%, respectively. The phantom was first simulated without attenuation (in air). Image intensities were estimated in Regions of Interest (ROIs) drawn in the reconstructed volumes for each cascade type as well as for their combination. Both cascade components contributed to an increase in spurious coincidences uniformly distributed throughout the image. A more realistic phantom (filled with water) was then simulated with or without modeling cascades. When cascade gamma rays are modeled, there is a slight increase of spurious activity in cold regions and a slight decrease of intensity in high-activity regions. While there is no significant change in LBR with or without cascade gamma events, the total counts increased greatly in the detector. To track the same number of beta(+), the total photon number is 4.4 times higher for Y-86 when modeling cascade events, causing a significant increase in detector dead time due to the increase of singles rate. The scatter fraction also increased, which was 18.95% without modeling cascades, and 26.75% when cascades were modeled. The simulation results are being validated with phantom experiments. Characterizing the spurious activity caused by cascade gamma rays in Y-86 PET imaging is crucial for the development of quantitative PET imaging of gamma ray emitting isotopes.