Patient-specific scatter correction in clinical cone beam computed tomography imaging made possible by the combination of Monte Carlo simulations and a ray tracing algorithm
Patient-specific scatter correction in clinical cone beam computed tomography imaging made possible by the combination of Monte Carlo simulations and a ray tracing algorithm
复制标题
DOI：
10.3109/0284186x.2013.813641
复制
发表时间：
2013-10-01
期刊：
影响因子：
3.1
通讯作者：
Brink, Carsten
中科院分区：
文献类型：
--
作者：
Thing, Rune S.;Bernchou, Uffe;Brink, Carsten
目的。圆锥束计算机断层扫描（CBCT）图像质量受到散射光子的限制。蒙特卡罗（MC）模拟提供了预测临床CBCT成像中患者特异性散射污染的能力。冗长的模拟会阻碍基于mc的散射校正在临床环境中完全实现。本研究探讨了使用快速MC模拟来预测散射分布与射线追踪算法的结合，以便在模拟和临床CBCT图像之间进行校准。材料和方法。使用基于egsnrc的用户代码（egs_cbct）对Elekta XVI CBCT成像系统进行MC模拟。采用60keV的x射线源，在探测面上进行空气雾化。为了提高散点计算效率，采用了多种方差减少技术（vrt）。基于CT扫描模拟了三个病人的幻象，即大脑、胸腔和骨盆扫描。采用射线追踪算法计算了探测器的主光子信号。总共模拟了288个投影，用于调查的计算机集群上每个线程一个。结果。脑、胸腔和骨盆扫描的散点分布在每次扫描两小时内的统计不确定性为2%。在同一时间内，光线追踪算法为每个投影提供了主要信号。因此，临床CBCT成像中基于mc的散射校正所需的所有数据都是在每位患者两小时内获得的，使用了临床CBCT几何结构的完全模拟。结论。本研究表明，在CBCT成像中使用基于mc的散射校正对提高CBCT图像质量具有很大的潜力。通过使用强大的vrt预测散射分布和光线追踪算法计算主信号，可以在每个患者两小时内获得患者特定MC散射校正所需的数据。
Purpose. Cone beam computed tomography (CBCT) image quality is limited by scattered photons. Monte Carlo (MC) simulations provide the ability of predicting the patient-specific scatter contamination in clinical CBCT imaging. Lengthy simulations prevent MC-based scatter correction from being fully implemented in a clinical setting. This study investigates the combination of using fast MC simulations to predict scatter distributions with a ray tracing algorithm to allow calibration between simulated and clinical CBCT images. Material and methods. An EGSnrc-based user code (egs_cbct), was used to perform MC simulations of an Elekta XVI CBCT imaging system. A 60keV x-ray source was used, and air kerma scored at the detector plane. Several variance reduction techniques (VRTs) were used to increase the scatter calculation efficiency. Three patient phantoms based on CT scans were simulated, namely a brain, a thorax and a pelvis scan. A ray tracing algorithm was used to calculate the detector signal due to primary photons. A total of 288 projections were simulated, one for each thread on the computer cluster used for the investigation. Results. Scatter distributions for the brain, thorax and pelvis scan were simulated within 2% statistical uncertainty in two hours per scan. Within the same time, the ray tracing algorithm provided the primary signal for each of the projections. Thus, all the data needed for MC-based scatter correction in clinical CBCT imaging was obtained within two hours per patient, using a full simulation of the clinical CBCT geometry. Conclusions. This study shows that use of MC-based scatter corrections in CBCT imaging has a great potential to improve CBCT image quality. By use of powerful VRTs to predict scatter distributions and a ray tracing algorithm to calculate the primary signal, it is possible to obtain the necessary data for patient specific MC scatter correction within two hours per patient.