自适应放疗中CT和信息缺失CBCT图像快速变形配准方法的研究

CT和CBCT图像变形配准是临床自适应放射治疗的一项关键技术，其精度直接关系到病人受照剂量的准确性。目前，CT-CBCT变形配准面临的两个主要问题是：CT和CBCT图像的CT值强度不一致以及CBCT使用小扫描野扫描引起的信息缺失，这两者均可能造成严重误配准。申请人的前期研究针对第一问题，建立了DISC模型，在校正CBCT图像CT值强度的同时可完成CT-CBCT变形配准。本项目在此基础上，提出了一种基于图形处理器GPU的CT与信息缺失CBCT图像快速变形配准新方法。研究内容：（1）构建及求解联合迭代重建与配准的能量函数，将CBCT全部投影图像信息以迭代前向投影与反投影的形式融入变形配准模型，同时完成CT与CBCT在投影域和图像域的匹配；（2）实现基于GPU的快速前向投影和反投影及其与DISC模型的融合；（3）模体及临床数据验证。本研究的开展对提高放射治疗计划重设计精度具有重要的临床应用价值。

CT to Cone-beam CT (CBCT) deformable image registration (DIR) plays an important role in adaptive radiation therapy (ART) to establish a correspondence between voxels in the CT and the CBCT for purposes such as transferring the organ contours from the planning CT images to the daily CBCT images for re-planning and dose re-optimization. It is hence desirable to have an accurate and robust DIR algorithm to facilitate this step. Conventional DIR algorithms may not be able to handle the DIR problem between CT and CBCT mainly because of two reasons: intensity inconsistency between CT and CBCT images and CBCT truncation which is caused by limited FOV(field of view) scan. In our previous work, we have proposed and implemented a modified demons algorithm embedded with a simultaneous intensity correction step, called deformation with intensity simultaneously corrected (DISC). DISC is immune to the CBCT artifacts and intensity inconsistency and can successfully register CT and CBCT images with CBCT intensity corrected simultaneously. Based on DISC, this project proposes a fast GPU (graphic process unit) based DIR algorithm for CT and truncated CBCT image, which can utilize all the CBCT projection information and incorporate them into an iterative forward/backward projection scheme with the DISC algorithm, it will potentially solve the DIR problem between CT and truncated CBCT images. This research mainly includes: (1) establishing a hybrid construction and registration objective function; (2) implementation of the fast forward/backward projection based on GPU architecture; (3) registration accuracy evaluation using a deformation phantom and clinical patient data.