Detection and correction for EPID and gantry sag during arc delivery using cine EPID imaging
Detection and correction for EPID and gantry sag during arc delivery using cine EPID imaging
复制标题
DOI：
10.1118/1.3673958
复制
发表时间：
2012-02-01
期刊：
影响因子：
3.8
通讯作者：
Greer, Peter B.
中科院分区：
文献类型：
--
作者：
Rowshanfarzad, Pejman;Sabet, Mahsheed;Greer, Peter B.
目的：电子射野成像设备（EPID）已被研究并用于预处理和体内剂量测定应用多年。在电弧治疗中应用EPID进行剂量测定需要准确表征旋转期间EPID和机架的机械下垂。有几项研究调查了重力对这些系统垂度的影响，但每项研究都有局限性。在这项研究中，一个简单的实验装置和准确的算法已经被引入到表征和纠正的效果EPID和机架弧delivery.Methods下垂：三个金属球轴承被用作标记的光束：其中两个固定在机架头和第三个定位在等中心。在电影成像模式下，在360度机架旋转期间采集EPID图像。在EPID图像中跟踪标记，并使用内部开发的强大MATLAB代码分析图像，并通过与参考机架零图像进行比较，找到三个方向上的EPID下垂以及EPID +机架下垂。然后对独立方法测试算法结果。该方法被应用于比较顺时针和逆时针机架旋转和不同的源-探测器距离（SDD）的效果。在15个月的时间内对一台直线加速器的结果进行了监测，并使用这种方法对来自两个治疗中心的其他六台直线加速器进行了调查。从数据中推导出广义移位模式，并将其用于图像配准算法中，以校正系统中机械下垂的影响。Gamma评价（3%，3 mm）技术被用于研究固定视野的电影EPID图像对准的改进，通过将单个图像和一系列图像的总和与参考机架零图像进行比较。机架旋转期间的机械下垂取决于机架角度，并且在平面内方向上较大，尽管对于各种线性加速器，图案并不相同。在15个月的时间内，测量的再现性在0.2 mm以内。机架旋转方向和SDD对结果的影响不超过0.3 mm。在测量工具的准确度范围内，独立测试的结果与算法一致。当比较总和图像时，Gamma指数< 1的点的百分比从85.4%增加到94.1%，校正后的EPID下垂，和99.3%校正后机架+ EPID sag.Conclusions：在这项研究中介绍的测量方法和算法使用电影图像，是高度准确的，简单，快速，可重复的。它测试所有机架角度，并提供合适的自动分析和校正工具，以改善EPID剂量测定，并为电弧治疗执行全面的直线加速器QA。(C)2012年美国医学物理学家协会。[DOI：10.1118/1.3673958]
Purpose: Electronic portal imaging devices (EPIDs) have been studied and used for pretreatment and in-vivo dosimetry applications for many years. The application of EPIDs for dosimetry in arc treatments requires accurate characterization of the mechanical sag of the EPID and gantry during rotation. Several studies have investigated the effects of gravity on the sag of these systems but each have limitations. In this study, an easy experiment setup and accurate algorithm have been introduced to characterize and correct for the effect of EPID and gantry sag during arc delivery.Methods: Three metallic ball bearings were used as markers in the beam: two of them fixed to the gantry head and the third positioned at the isocenter. EPID images were acquired during a 360 degrees gantry rotation in cine imaging mode. The markers were tracked in EPID images and a robust in-house developed MATLAB code was used to analyse the images and find the EPID sag in three directions as well as the EPID + gantry sag by comparison to the reference gantry zero image. The algorithm results were then tested against independent methods. The method was applied to compare the effect in clockwise and counter clockwise gantry rotations and different source-to-detector distances (SDDs). The results were monitored for one linear accelerator over a course of 15 months and six other linear-accelerators from two treatment centers were also investigated using this method. The generalized shift patterns were derived from the data and used in an image registration algorithm to correct for the effect of the mechanical sag in the system. The Gamma evaluation (3%, 3 mm) technique was used to investigate the improvement in alignment of cine EPID images of a fixed field, by comparing both individual images and the sum of images in a series with the reference gantry zero image.Results: The mechanical sag during gantry rotation was dependent on the gantry angle and was larger in the in-plane direction, although the patterns were not identical for various linear-accelerators. The reproducibility of measurements was within 0.2 mm over a period of 15 months. The direction of gantry rotation and SDD did not affect the results by more than 0.3 mm. Results of independent tests agreed with the algorithm within the accuracy of the measurement tools. When comparing summed images, the percentage of points with Gamma index < 1 increased from 85.4% to 94.1% after correcting for the EPID sag, and to 99.3% after correction for gantry + EPID sag.Conclusions: The measurement method and algorithms introduced in this study use cine-images, are highly accurate, simple, fast, and reproducible. It tests all gantry angles and provides a suitable automatic analysis and correction tool to improve EPID dosimetry and perform comprehensive linac QA for arc treatments. (C) 2012 American Association of Physicists in Medicine. [DOI: 10.1118/1.3673958]