基于感兴趣窄带区域的同步分割与配准方法及在IGRT中的应用

Synchronous Segmentation and Registration Method Based on Narrow Band of Interest and Its Application to IGRT System

医学图像分割与配准是图像引导放疗(Image guided radiation therapy,IGRT)系统中的关键技术.为提高基于CBCT(Cone beam CT)的IGRT系统实施胸腹部肿瘤放疗的实时性与自适应性,特别是实现重要危及器官肝脏区域照射剂量的合理控制,本文提出一种基于感兴趣窄带区域的同步分割与配准方法,目标是实现放疗计划系统中计划CT和CBCT图像目标区域的分割与配准.通过构建感兴趣窄带模型,并且与活动轮廓模型相结合实现初始分割,然后与基于光流场(Optical flow field,OFF)的形变配准方法进行循环迭代,从而构造ASOR分割与配准同步模型(Active contour segmentation and optical flow registration synchronously,ASOR).在方法实施时,首先利用非线性扩散模型和窄带活动轮廓模型在CT图像中提取肝脏空间初始位置信息,为同步模型提供合理的肝脏初始轮廓.然后将该轮廓及相应窄带区域经仿射变换映射到CBCT图像,进而结合构造的ASOR同步模型,用光流场确定活动轮廓水平集的运动情况,使分割与配准在同一个演化过程中完成迭代.实验结果和临床应用表明,本文提出的方法应用于基于CBCT的IGRT系统时,可实现肝脏组织的自动分割与放疗剂量分布的快速计算.同时,我们将同步过程中获得的形变域用于实现肝脏与肿瘤靶区等剂量线从计划CT到CBCT的自适应转移,进行自适应放疗效果的临床测评.

Medical image segmentation and registration is a key technology in image guided radiation therapy（IGRT）system. In order to improve the real-time performance of cone beam CT（CBCT） based IGRT system for thoracic and abdominal tumors treatment, also for controlling projection dose in liver area efficiently, a synchronous segmentation and registration joint method based on narrow band of interest is proposed to achieve segmentation and registration for the radiotherapy treatment planning system. The key issue in our method is to construct an ASOR synchronization model by integrating narrow band model with active contour model to accomplish initial segmentation, then is combined with an optical flow based deformable registration method to optimize the process iteratively. At first, both nonlinear diffusion model and narrow band active contour are used to get the liver position information of the CT image to provide reasonable initial contour for the synchronization model. Secondly, the liver contour and corresponding narrow band are mapped from the planning CT to CBCT by affine transformation. Thirdly and finally, the ASOR synchronization model is used to fulfill segmentation and registration simultaneously in evolution process by optical flow for determining the active contour level set movements. The experiment results demonstrate that when the proposed method is applied to CBCT based IGRT system, it can automatically segment liver to implement the real time calculation for the following radiation therapy planning, and that the deformation field which is obtained during the segmentation process can transfer the radiation planning from planning CT to CBCT adaptively.