Faculty of Physics, Sofia University, Sofia, Bulgaria.
Med Phys. 2012 Jul;39(7):4175-86. doi: 10.1118/1.4725168.
The purpose of this work is to study the influence of photon energy cuts on the results of positron emission tomography (PET) Monte Carlo (MC) simulations.
MC simulations of PET scans of a box phantom and the NEMA image quality phantom are performed for 32 photon energy cut values in the interval 0.3-350 keV using a well-validated numerical model of a PET scanner. The simulations are performed with two MC codes, egs_pet and GEANT4 Application for Tomographic Emission (GATE). The effect of photon energy cuts on the recorded number of singles, primary, scattered, random, and total coincidences as well as on the simulation time and noise-equivalent count rate is evaluated by comparing the results for higher cuts to those for 1 keV cut. To evaluate the effect of cuts on the quality of reconstructed images, MC generated sinograms of PET scans of the NEMA image quality phantom are reconstructed with iterative statistical reconstruction. The effects of photon cuts on the contrast recovery coefficients and on the comparison of images by means of commonly used similarity measures are studied.
For the scanner investigated in this study, which uses bismuth germanate crystals, the transport of Bi X(K) rays must be simulated in order to obtain unbiased estimates for the number of singles, true, scattered, and random coincidences as well as for an unbiased estimate of the noise-equivalent count rate. Photon energy cuts higher than 170 keV lead to absorption of Compton scattered photons and strongly increase the number of recorded coincidences of all types and the noise-equivalent count rate. The effect of photon cuts on the reconstructed images and the similarity measures used for their comparison is statistically significant for very high cuts (e.g., 350 keV). The simulation time decreases slowly with the increase of the photon cut.
The simulation of the transport of characteristic x rays plays an important role, if an accurate modeling of a PET scanner system is to be achieved. The simulation time decreases slowly with the increase of the cut which, combined with the accuracy loss at high cuts, means that the usage of high photon energy cuts is not recommended for the acceleration of MC simulations.
本工作旨在研究光子能量截断对正电子发射断层扫描（PET）蒙特卡罗（MC）模拟结果的影响。
使用经过充分验证的 PET 扫描仪数值模型，对 32 个光子能量截断值在 0.3-350keV 范围内的盒式 phantom 和 NEMA 图像质量 phantom 的 PET 扫描进行 MC 模拟。使用两种 MC 代码，即 egs_pet 和 GEANT4 Application for Tomographic Emission（GATE）进行模拟。通过将较高截断值的结果与 1keV 截断值的结果进行比较，评估光子能量截断对记录的单光子、初级、散射、随机和总符合计数以及模拟时间和噪声等效计数率的影响。为了评估截断对重建图像质量的影响，使用迭代统计重建对 NEMA 图像质量 phantom 的 PET 扫描的 MC 生成的正弦图进行重建。研究了光子截断对对比度恢复系数的影响，以及使用常用相似性度量标准对图像进行比较的影响。
对于本研究中使用锗酸铋晶体的扫描仪，必须模拟 Bi X（K）射线的输运，才能对单光子、真实、散射和随机符合计数以及噪声等效计数率进行无偏估计。高于 170keV 的光子能量截断会导致康普顿散射光子的吸收，并大大增加所有类型的记录符合计数和噪声等效计数率。光子截断对重建图像及其用于比较的相似性度量的影响在非常高的截断值（例如，350keV）时具有统计学意义。随着光子截断的增加，模拟时间缓慢减少。
如果要实现对 PET 扫描仪系统的精确建模，则特征 X 射线的输运模拟起着重要作用。随着截断值的增加，模拟时间缓慢减少，而在高截断值时的精度损失意味着不建议使用高光子能量截断来加速 MC 模拟。
