Metal Artifact Reduction (MAR) in Image-Guided Radiation Therapy (IGRT)

x射线计算机断层扫描（CT）的图像质量经常受到金属植入物产生的伪影的影响。例如，这些人工制品可能是由于假体，由于牙齿填充物，由于固定物，夹子或标记。金属伪影可能会损害诊断，或者它们可能会损害放射治疗计划所需的图像定量评估。金属伪影的类型和强度取决于金属植入物的类型、大小、密度、位置和数量，以及CT扫描的扫描参数。其他因素，如有意或无意的患者运动也起着重要作用。这尤其适用于图像引导放射治疗（IGRT），其中使用扁平探测器锥束CT技术在自由呼吸下获取患者数据，该技术比诊断CT系统更容易产生伪影。目前的方法不能减少或消除这种移动的金属伪影，因为它们出现在IGRT中。文献中描述了几种金属伪影减少（MAR）的方法。其中一些用一些替代信息代替穿过金属物体的射线的投影值。另一些则不是基于这种涂漆技术，而是利用金属阴影中的信息和测量过程的物理模型来获得校正的原始数据。在大多数情况下，磁共振成像用于临床（诊断）CT。尽管平面探测器CT系统对散射辐射更敏感，因此对金属制品也更敏感，但专门针对该系统的研究较少。除了一种手动方法外，目前还不知道专门用于图像引导放射治疗的方法，尽管IGRT必须满足特殊要求，才能使治疗计划适应当前患者的情况。一方面，图像质量必须高，以便能够可靠地描绘肿瘤边缘。例如，在前列腺的情况下，周围的髋关节假体可能会产生金属伪影，延伸到目标体积，并可能阻止治疗计划的准确适应。在其他情况下，肿瘤用金属标记物标记，同时也产生金属伪影。另一方面，自由的病人呼吸是这些扫描的典型特征。然而，患者的运动改变或放大了金属伪影，使金属物体变得模糊。然而，这种模糊可能会妨碍植入物的可靠分割，但这种分割是减少金属伪影的先决条件。该项目旨在开发一种用于图像引导放射治疗的专用MAR算法。一方面，我们要考虑平面探测器的特殊特性，并希望开发一种专用的运动补偿方法，使MAR也能应用于移动金属的情况。可行性应通过模拟、模拟测量和患者测量来证明。

Image quality in x-ray computed tomography (CT) often suffers from artifacts arising from metal implants. For example these artifacts can be due to prostheses, due to dental fillings, due to fixations, clips or markers. The metal artifacts may impair diagnosis or they may impair quantitative evaluations of the images such as it is required for radiation treatment planning. The type and intensity of the metal artifacts depends on the type, size, density, position and number of the metal implants, and on the scan parameters of the CT scan. Additional factors such as intended or unintended patient motion play a significant role, too. This particularly applies for image-guided radiation therapy (IGRT) where patient data are acquired under free breathing using flat detector cone-beam CT technology which is much more prone to artifacts than diagnostic CT systems. Current methods are not able to reduce or remove such moving metal artifacts, as they arise in IGRT.Several methods for metal artifact reduction (MAR) are described in the literature. Some of them replace the projection values of rays that run through a metal object by some surrogate information. Others are not based on such inpainting techniques but rather use the information in the metal shadow and a physical model of the measurement process to come up with corrected rawdata. In most cases MAR is dedicated to clinical (diagnostic) CT. Less work has been published dedicated to flat detector CT systems, although those are much more sensitive to scattered radiation and therefore to metal artifacts. Approaches dedicated to image-guided radiation therapy that cope with patient motion are not known, apart from one manual method, although IGRT has to fulfill special requirements to be able to adapt the treatment plan to the current patient situation. On the one hand image quality should be high to be able to reliably delineate the tumor margins. In case of the prostate, for example, surrounding hip prostheses may generate metal artifacts that extend into the target volume and that potentially prevent an accurate adaptation of the treatment plan. In other cases tumors are marked with metal markers that also generate metal artifacts. On the other hand free patient breathing is typical for those scans. Patient motion, however, changes or amplifies the metal artifacts and blurs the metal object. This blurring, however, may prevent a reliable segmentation of the implants but this segmentation is a prerequisite for metal artifact reduction.This project aims at developing a dedicated MAR algorithm for image-guided radiation therapy. We want to consider the particular properties of flat detectors on the one hand, and want to develop a dedicated motion compensation approach that allows to apply MAR also to the case of moving metal. The feasibility shall be demonstrated using simulations, phantom measurements, and patient measurements.