摘要
目的筛选、评价适合应用于千伏特锥形束CT（KVCBCT）在线引导肺癌放疗的图像配准方法。方法选择16例行根治性放疗的非小细胞肺癌患者进入研究，每例患者每周行KVCBCT在线引导体位校正1次，共96幅KVCBCT图像用于研究。分别采用基于灰度的自动配准法、骨性结构的自动配准法、人工图像配准法和自动加手动图像配准法配准患者KVCBCT图像与计划设计CT图像。配准由1名医生完成，配准结果由另外1名医生在对所用配准方法设盲的情况下进行评价与筛选，并对筛选出的最优图像配准方法的可重复性进行评价。结果基于灰度的自动配准法、骨性结构的自动配准法、人工图像配准法和自动加手动图像配准法的平均得分分别为2．7、2．4、3．0和3．7分，4个组评分差异有统计学意义（F=42．20，P〈0．001）。采用自动加手动图像配准法左右（LR）、头脚（SI）和前后（AP）方向上，同一医生2次配准结果差值〉3mm所占比例分别为0、3％和6％，不同医生的分别为0、14％和0，医生与技术员的分别为8％、14％和8％。结论自动加手动图像配准法以其配准结果好、所需时间较短、临床应用可重复性强等特点适合应用于肺癌KVCBCT在线引导放疗。
Objective To select the optimal registration method for on-line kilovoltage cone-beam CT （KVCBCT） guided lung cancer radiation and evaluate the reproducibility of the selected method. Methods Sixteen patients with non-small cell lung cancer were enrolled into this study. A total of 96 pretreatment KVCBCT images from the 16 patients were available for the analysis. Image registration methods were bone-based automatic registration, gray-based automatic registration, manual registration and semi-automatic registration. All registrations were accomplished by one physician. Another physician blindly evaluated the results of each registration, then selected the optimal registration method and evaluated its reproducibility. Results The average score of the bone-based automatic registration, gray-based automatic registration, manual registration and semi-automatic registration methods was 2.4,2.7,3.0 and 3.7, respectively. The score of the four different groups had statistics significant difference （ F = 42.20, P 〈 0.001 ）. Using the semi-automatic registration method, the probability of the difference between two registration results more than 3 mm in the left-right, superior-inferior, and anterior-posterior directions was 0,3% and 6% by the same physician,0,14% and 0 by different physicians, and 8% , 14% and 8% by physician and radiation therapist. Conclusions Semi-automatic registration method, possessing the highest score and accepted reproducibility, is appropriate for KVCBCT guided lung cancer radiation.
出处
《中华放射肿瘤学杂志》
CSCD
北大核心
2009年第1期61-64,共4页
Chinese Journal of Radiation Oncology
基金
基金项目：卫生部临床学科重点项目（2004468）
上海市科委重点项目（34119814）
关键词
图像引导放疗
锥形束CT
图像配准
肺肿瘤
Image-guided radiotherapy
Cone-beam CT
Image registration
Lung neoplasm
