摘要目的：比较肝癌千伏级 CBCT 图像与计划 CT 图像配准的不同方法。方法离线下由4位观测者在加速器控制室内分别完成121幅 CBCT 图像与计划 CT 图像的配准，配准方法包括手动配准、常规软组织配准、常规骨性配准、肝脏配准、椎体配准，其中后4种方法均使用配准软件自动配准。对配准软件的稳定性进行检测，然后比较4位观测者使用同种自动配准方法结果，以此评估使用该方法可重复性。以4位观测者手动配准肝脏外轮廓结果平均值为基准，与其他配准方法结果比较。使用 McNemar 检验差异。结果若保持配准条件不变，自动配准结果完全一致。常规软组织、骨性配准方法可重复性较好，椎体配准次之，肝脏配准最差。但肝脏配准结果与手动配准结果最接近，差值在3 mm 内百分率分别为左右方向84??3％、上下方向77??7％、前后方向92??6％。结论使用配准软件自动肝脏配准可用于肝癌图像引导，但每次图像引导时均需有经验医师、技师共同参与，自动配准完成之后应根据肝脏外轮廓手动校正。

abstractsObjective To study the methods of matching kilovolt CBCT image with planning CT scan. Methods A total of 121 CBCT scans were matched with planning CT scan using a manual and four automatic match methods by four observers in the offline. In the manual match, the live contour was used as a surrogate for image registration. Four automatic match methods, including routine soft?tissue match, routine bone match, automatic liver match and vertebral body match, were performed using image registration sofeware. First, the stability of the sofeware was tested. Then, the reproducibility of the same automatic match method was evaluated by comparing different observers’ match results. After the manual match by four observers, the mean of the match results was used as a standard to compare with others. The differerces was test by McNemar method. Results In the uniform match factors, automatic match result would not change. The reproducibility of routine soft?tissue and bone match are best, automatic vertebral body match is better than automatic liver match. Howerver, the automatic liver match result is the most similar to manual match, the percentages of match result have an absolute error no more than 3 mm in left?right, superior?inferior and anterior?posterior directions are 84?? 3%, 77?? 7% and 92?? 6%. Conclusions Automatic liver match can be used in image?guided radiotherapy for liver cancer, however, it should be performed by experienced oncologist and technologist together in each fraction, after that, the matching result should be adjusted carefully according to live contour.