Dipartimento di Fisiopatologia Clinica, Universita' di Firenze, 50134 Firenze, Italy.
J Appl Clin Med Phys. 2010 Aug 25;11(4):3180. doi: 10.1120/jacmp.v11i4.3180.
In radiotherapy, treatment portal images and digitally reconstructed radiographs (DRRs) are used to monitor patient setup during clinical routine. The output of the registration between the portal image and the reference DRR indicates the patient displacement. If the registration is not reliable, the patient positioning will not be accurate. The aim of this work is to assess the intrinsic and the global accuracy of iView and PIPSpro, two widely used registration software programs that implement a manual and a semiautomatic approach, respectively. The intrinsic accuracy was tested using a computer generated phantom, while the overall accuracy was evaluated registering the portal images and the DRRs of an Alderson RANDO phantom. For DRRs, four treatment planning systems (TPS) and three CT studies with different slice thicknesses were considered.This study demonstrates that the intrinsic accuracy of iView and PIPSpro were within 1 pixel and 1°. Using a DRR extracted from a 2 mm CT study, the overall accuracy of both methods was about 2 mm and 1°. When thicker CT slices are considered, the global accuracy of both methods worsens, and differences larger than 1.5° between the rotation parameters estimated with iView and PIPSpro are evident. The results obtained with iView and PIPSpro were nearly equivalent.
在放射治疗中,治疗门户图像和数字重建射线照片(DRR)用于在临床常规中监测患者的设置。门户图像和参考 DRR 之间的配准的输出指示患者的位移。如果配准不可靠,则患者的定位将不准确。本工作的目的是评估两种广泛使用的配准软件程序 iView 和 PIPSpro 的固有和全局精度,它们分别实现了手动和半自动方法。固有精度使用计算机生成的幻影进行测试,而整体精度则通过注册门户图像和 Alderson RANDO 幻影的 DRR 进行评估。对于 DRR,考虑了四个治疗计划系统(TPS)和三个具有不同切片厚度的 CT 研究。这项研究表明,iView 和 PIPSpro 的固有精度在 1 像素和 1°以内。使用来自 2mm CT 研究的 DRR,两种方法的整体精度均约为 2mm 和 1°。当考虑更厚的 CT 切片时,两种方法的整体精度都会恶化,并且在 iView 和 PIPSpro 中估计的旋转参数之间存在大于 1.5°的差异。iView 和 PIPSpro 的结果几乎相等。