【作者】 尹勇；

【导师】 万柏坤；

【作者基本信息】 天津大学 ， 生物医学工程， 2009， 博士

【摘要】 肿瘤放射治疗技术已进入以影像引导的三维适形放疗和调强放疗为代表的精确放疗新时代。影像引导的放射治疗(IGRT)技术的开展,尤其是锥形束CT(CBCT)成像系统的使用,大大提高了放射治疗的精度。目前基于CBCT成像的IGRT研究在国内外方兴未艾,其临床应用尚存在若干关键问题急待解决。IGRT临床应用的主要关键问题之一是要对CBCT重建靶区影像与治疗计划CT进行精确快速的配准。目前的配准技术在速度和精度上尚难达到临床实施要求。本论文采用了基于互信息的多分辨率配准算法框架以减少配准时间,从理论和实验两方面阐明了基于多尺度空间的三维医学图像多分辨率配准方案可以显著提高配准速度。完成了基于互信息刚性配准算法的CBCT影像重建三维病人模型与治疗计划CT病人模型配准实验,比较了多分辨率配准框架中高斯金字塔和直接抽取塔式结构的配准性能。研究表明,直接抽取结构框架可以更快地完成配准,且鲁棒性更好;基于互信息配准方法可以有效地克服目前常用基于骨形特征的有框架配准缺陷,具有精度高、鲁棒性好、无需提取特征、易实现全自动配准等优点,适合用于CBCT与计划CT的配准实验及指导摆位。IGRT临床应用尚待解决的另一关键是缺乏CBCT图像质量的评价标准及质量控制方法。论文对此进行了开创性研究,建立了以图像均匀性、高低分辨率、CT值稳定性、空间线形、焦点辐射剂量等多项参数为指标的CBCT图像质量分析评价标准、方法及质量控制体系,开拓了基于CBCT图像的放疗计划设计和评价新途径。论文研制了二维运动体模系统以探讨器官运动和形变对CBCT成像的影响。实验表明:呼吸运动在横轴和纵轴方向对体模CBCT成像均有明显影响,皆使图像变形、体积增大、清晰度下降,运动幅度越大,效果越显著;呼吸频率对图像大小影响不明显。为解决目前CBCT存在的扫描区域狭窄问题,论文设计了对分段锥形束CT图像进行拼接以获得完整图像的新方法。其过程为:将两段图像分别进行显像解析,再经三维重建后分别与计划CT进行图像融合,通过对融合后骨性标志的分析来选取拼接层面来实现两段CBCT图像的拼接处理。新方法扩大了肿瘤靶区及周围正常器官的观察范围,保证了基于CBCT图像制定放疗计划的可行性及评价剂量体积参数的完整性,有广泛的临床应用前景。更多还原

【Abstract】 Tumor radiation therapy has entered into a new era, in which tumor is being treated by 3-dimentional conformal radiotherapy (3D-CRT) and intensity modulated radiotherapy (IMRT). A typical representative of the new treatment technology is the image guided radiotherapy (IGRT) based on the cone-beam computed tomography (CBCT), with which there’s a great improvement on the accuracy and precision of treatment.One of the key issues in IGRT clinical application is achieving accurate and rapid registration for Planning CT and CBCT. Existing registration methods cannot still satisfy the clinical implementation requirements in the speed and accuracy. In this paper, mutual information based multi-resolution registration framework is used to accelerate the registration process. Theoretically and experimentally, it’s demonstrated that multi-scale space based multi-resolution three-dimensional medical image registration framework can greatly increase the speed of registration process rather to direct registration. We realized that mutual information based rigid registration algorithm for IGRT system by registering CBCT of the three-dimensional image reconstruction of patients with the Planning CT of the patient treatment model. Furthermore, two multi-resolution registration frameworks are compared: direct equidistant sampling pyramid and Gaussian pyramid. Studies have shown that direct equidistant sampling pyramid framework can be implemented quickly to complete registration, and with better robustness; and moreover mutual information-based registration method can be used to overcome the limitation of the current bone-shaped characteristics based framework for the registration. Now we can get high precision and robustness, and there is no need to extract features and easy to achieve automatic registration etc. It is very suitable for CBCT and Planning CT registration for image guided radiation therapy.Another key issue is the evaluation standard and quality control system on CBCT. The dissertation studied on and developed innovatively an evaluation standard and quality control system, which are based on the image uniformity, image resolution, CT value stability, space linearity, and dose on the isocenter. With them, it is more feasible and easy to design and evaluate a radiotherapy plan based on CBCT.A 2D kinetic phantom system, which is used to simulate respiratory motion, has been developed to study on the impacts of organs motion and deformation upon CBCT. From tests, it has been revealed that the impacts of respiratory motion upon the reconstruction and deformation of CBCT images in latitude and longitude directions were remarkable. Compared to the CBCT images under quiescence, it was fuzzy and deformed, with a larger volume. Respiratory extent played an important role in it, but the frequency not.The dissertation has studied and developed a new program to paste two respective CBCT images to enlarge length of CBCT images, and offer the integrity of structures. The process is that importing two CBCT image series to TPS and then fusion the planning CT and CBCT together. Find the same layer of two CBCT series by analyzing the two fusion results and the bone as the reference. When enlarged, more information including tumor and organs at risk (OAR) could be integrated. This new technique can provide the chance to observe the whole target and OAR and the chance for physicists to evaluating the treatment plan and re-planning based on CBCT.更多还原