肺癌放疗两种不同体位固定方式的差异性及摆位误差的相关性分析

目的　探究肺癌放疗两种不同体位固定方式的差异性及摆位误差的相关性。方法　回顾性分析2019年2月至2021年5月中山大学肿瘤防治中心接受放疗的肺癌患者108例（男84例，女24例，中位年龄58岁），所有患者均采用负压真空袋固定。按照固定方式不同分为A组（双手上举52例）、B组（双手下垂56例）。患者治疗前使用机载锥形束CT（CBCT）扫描，总共得到容积CT图像677次。重建后和定位CT图像进行六维配准，记录摆位误差数值。原始数据取绝对值后行Kolmogorov-Smirnov Z非参数秩和检验，并对原始数据进行相关性分析。结果　A组和B组六维度配准误差取绝对值后，在头脚（SI）、左右（LR）、腹背（AP）方向的误差值依次为：0.20（0.30）cm、0.20（0.20）cm、0.20（0.20）cm和0.20（0.30）cm、0.20（0.20）cm、0.10（0.20）cm；原始数据取绝对值后进行非参数秩和检验分析：Z值分别为0.875、0.844、2.164，AP方向比较差异有统计学意义（P=0.007）；A组和B组在冠状位（RTN）、矢状位（Pitch）、横断位（Roll）方向的误差依次为：0.50（0.60）°、0.60（0.70）°、0.70（0.90）°和0.50（0.60）°、0.80（1.20）°、0.60（1.00）°；原始数据取绝对值后进行非参数秩和检验分析：Z值分别为0.942、2.645、2.356，Pitch与Roll方向比较差异均有统计学意义（均P<0.05）。SI方向的误差与LR、AP、Pitch方向的误差有相关，相关系数r值分别为0.209、0.338、0.287，差异均有统计学意义（均P<0.01）；AP方向的误差与Pitch、Roll方向的误差有相关，相关系数r值分别为0.150、-0.086，差异均有统计学意义（均P<0.05）。结论　线性方向：B组在AP方向的摆位误差比A组小；旋转方向：A组在Pitch方向的摆位误差比B组小，B组在Roll方向的摆位误差比A组小。根据相关性分析，SI方向误差与LR、Pitch方向误差低度相关，与AP方向误差中度相关；AP方向误差与Pitch、Roll方向误差低度相关。临床应用可根据实际情况选取不同的固定方式。

Differences between two different postural fixation methods and correlations of setup errors in radiotherapy for lung cancer

Objective　To explore the differences between two different postural fixation methods and the correlations of setup errors in radiotherapy for lung cancer. Methods　A retrospective analysis was performed on 108 patients with lung cancer who received radiotherapy in Sun Yat-sen University Cancer Center (84 males and 24 females, with a median age of 58 years old). All patients were fixed with negative pressure vacuum bags. According to the different fixation methods, they were divided into group A (52 cases with hands up) and group B (56 cases with hands down). The patients were scanned with airborne cone beam CT (CBCT) before treatment, and a total of 677 volume CT images were obtained. Six-dimensional registration was performed on the reconstructed and localized CT images, and the setup error values were recorded. The Kolmogorov-Smirnov Z nonparametric rank sum test was performed after taking the absolute value of the original data, and the correlation analysis was performed on the original data. Results　After taking the absolute value of the six-dimensional registration errors in group A and B, the error values in the left-right (LR), superior-inferior (SI), and anterior-posterior (AP) directions were successively as follows: 0.20 (0.30) cm, 0.20 (0.20) cm, 0.20 (0.20) cm, and 0.20 (0.30) cm, 0.20 (0.20) cm, 0.10 (0.20) cm; after taking the absolute value of the original data, the nonparametric rank sum test analysis was performed: Z values in the SI, LR, and AP directions were 0.875, 0.844, and 2.164, respectively, and there was a statistically significant difference in the AP direction (P=0.007). The errors in the RTN, Pitch, and Roll directions of group A and B were 0.50 (0.60)°, 0.60 (0.70)°, 0.70 (0.90)°, and 0.50 (0.60)°, 0.80 (1.20)°, 0.60 (1.00)°; after taking the absolute value of the original data, the nonparametric rank sum test analysis was performed: Z values were 0.942, 2.645, and 2.356, respectively, and there were statistically significant differences in the Pitch and Roll directions (both P<0.05). The error in the SI direction was correlated with the errors in the LR, AP, and Pitch direction, the correlation coefficient r values were 0.209, 0.338, and 0.287, respectively, and the differences were statistically significant (all P<0.01). The error in the AP direction was correlated with the errors in the Pitch and Roll direction, the correlation coefficient r values were 0.150 and -0.086, respectively, and the differences were statistically significant (both P<0.05). Conclusions　Linear direction: the setup error of group B in the AP direction was smaller than that of group A; rotation direction: the setup error of group A in the Pitch direction was smaller than that of group B, and the setup error of group B in the Roll direction was smaller than that of group A. According to the correlation analysis, the error in the SI direction was low correlated with the errors in the LR and Pitch direction, and moderately correlated with the error in the AP direction; the error in the AP direction was low correlated with the error in the Pitch and Roll direction. In clinical application, different fixation methods could be selected according to the actual situation.