Four Dimensional Cone Beam CT for Image-guided Radiation Therapy

描述（由申请人提供）：近年来，在许多临床中心已经开发了利用诊断CT扫描仪和替代呼吸信号的四维CT （4DCT）成像方案。利用不同呼吸相的4DCT图像体积，提取呼吸运动场，从而估计肿瘤的运动轨迹。提取的运动场可以用来在给定的参考呼吸阶段扭曲图像体积和治疗方案。然而，运动场和运动轨迹可能会经历从治疗计划时间到治疗交付时间的变化。因此，开发一种方法来监测运动轨迹的潜在变化是很重要的，这既可以在实际治疗之前验证治疗计划，也可以为治疗重新计划获得必要的信息。这促使研究人员将4DCT的概念和方法扩展到4D锥束CT （4DCBCT）病例，其中获得的锥束投影使用替代呼吸信号分类为不同的呼吸期。然而，4DCBCT存在两个根本性的挑战，阻碍了其图像质量的提高，从而阻碍了其应用。第一个挑战是严重违反了Shannon/Nyquist采样要求。当获得的600个投影被门控为8-10个呼吸相时，只有60-80个锥束投影可用于重建每个呼吸相。当标准的图像重建算法，如滤波后的反向投影（FBP），应用于高度欠采样的数据集时，条纹伪影在重建图像中猖獗。4DCBCT的第二个挑战是给定呼吸期的锥束投影数据的采样模式远非最佳。这使得可用的60-80个投影的使用效率大大降低。因此，目前的4DCBCT图像质量不足以在常规临床实践中用于提取准确的运动轮廓，以保证质量或重新规划治疗目的。本提案的总体目标是开发一种创新的方法来提高4DCBCT图像质量，并展示其在图像引导放射治疗中的应用。核心使能技术是PI小组新提出的图像重建方案，即先验图像约束压缩感知（PICCS）。使用这种新方法，初步结果表明，CT图像可以使用大约10-20个投影精确重建，这使得获得的锥束投影可以门控到20-30个呼吸相。因此，4DCBCT成像可以使用60秒的数据采集来实现，而不是延长4-5分钟的扫描时间。在本提案中，中心假设是PICCS采集和图像重建方法应该能够重建用于放射治疗应用的高时间分辨率和无条纹的4DCBCT图像。这种新方法被称为PICCS-4DCBCT。本文的具体目标是：(1)优化PICCS图像重建算法的实现；(2)制定和评估PICCS-4DCBCT数据采集协议；(3)应用PICCS-4DCBCT进行临床评价。项目完成后，使用新型PICCS图像重建方法的精确4DCBCT将被开发并验证用于图像引导放射治疗。这项新技术将能够准确地提取肿瘤运动信息，以补偿治疗前和治疗过程中与呼吸相关的抽吸和/或干涉运动。

DESCRIPTION (provided by applicant): In recent years, four-dimensional CT (4DCT) imaging protocols utilizing a diagnostic CT scanner and surrogate respiratory signal have been developed at many clinical centers. Using 4DCT image volumes at different respiratory phases, respiratory motion fields can be extracted, thereby allowing estimation of motion trajectory of the tumor. The extracted motion fields can be utilized to warp the image volume and treatment plan at a given reference respiratory phase. However, the motion fields and the motion trajectory may experience changes from the time of treatment planning to the time of treatment delivery. Thus, it is important to develop a method to monitor the potential changes to the motion trajectory to either verify the treatment plan before the actual treatment is delivered or obtain necessary information for treatment replanning. This motivated investigators to extend the 4DCT concept and methodology to the 4D cone-beam CT (4DCBCT) case, in which the acquired cone-beam projections are sorted into different respiratory phases using surrogate respiratory signals. However, there are two fundamental challenges in 4DCBCT that hinder the improvement of its image quality, and thus its applications. The first challenge is a significant violation of the Shannon/Nyquist sampling requirement. When the acquired 600 projections are gated into 8-10 respiratory phases, there are only 60-80 cone-beam projections available to reconstruct each respiratory phase. When standard image reconstruction algorithms, such as Filtered BackProjection (FBP), are applied to highly undersampled data sets, streaking artifacts are rampant in the reconstructed image. The second challenge in 4DCBCT is the sampling pattern of the cone-beam projection data for a given respiratory phase is far from optimal. This makes the use of the available 60-80 projections significantly inefficient. As a consequence, the current 4DCBCT image quality is insufficient to be utilized in routine clinical practice to extract accurate motion profiles either for quality assurance or for treatment replanning purposes. The overall objective of this proposal is to develop an innovative method to improve the 4DCBCT image quality and to demonstrate its use in image-guided radiation therapy. The central enabling technology is a newly proposed image reconstruction scheme by the PI's group, namely Prior Image Constrained Compressed Sensing (PICCS). Using this novel method, preliminary results have demonstrated that CT images can be accurately reconstructed using approximately 10-20 projections, which enables gating of the acquired cone- beam projection into 20-30 respiratory phases. As a result, 4DCBCT imaging can be achieved using a 60- second data acquisition, rather than a prolonged 4-5 minute scan time. In this proposal, the central hypothesis is that the PICCS acquisition and image reconstruction method should enable reconstruction of high temporal resolution and streak-free 4DCBCT images for radiation therapy applications. This new method is referred to as PICCS-4DCBCT. The specific aims of the proposal are to: (1) Optimize the implementation of the PICCS image reconstruction algorithm; (2) Develop and evaluate PICCS-4DCBCT data acquisition protocols; (3) Conduct clinical evaluations using PICCS-4DCBCT. Upon the completion of the project, accurate 4DCBCT using the novel PICCS image reconstruction method will have been developed and validated for image-guided radiation therapy. This new technology will enable accurate extraction of tumor motion information for compensation of the respiratory-related intrafraction and/or interfraction motion before and during treatment delivery. PUBLIC HEALTH RELEVANCE: As advances in radiotherapy technology have enabled treatment with increasing precision, they have underscored the importance of geometrical uncertainties. The increasingly conformal dose distributions are more sensitive to uncertainties that, if not dealt with properly, may result in tumor underdose or normal tissue overdose. A prerequisite for highly precise dose delivery is highly precise targeting. Any motion of the patient or of the inner organs either during or between the treatment fractions, as well as any setup errors, can have a detrimental effect on the outcome of the treatment. In this proposal, we plan to develop a 4DCBCT method using the on-board cone-beam CT to accurately track the tumor motions caused by the respiratory motions. This will find wide application in cancer treatment for the chest and upper abdomen, including cancer of either the lungs or liver.