目的 评估光学体表监测系统（optical surface monitoring system，OSMS）在胸部肿瘤调强放疗摆位中的应用，并分析其与体质量指数（BMI）的关系。 方法 随机选取49例接受常规调强放疗的胸部肿瘤患者，分别获取有体膜时CBCT与定位CT的摆位误差及无体膜时OSMS与首次体表参考影像的配准误差。应用配对t检验分析两种图像引导方式配准误差的差异性，Pearson法分析CBCT误差和OSMS误差的相关性，Bland-Altman法评估两种误差的一致性。分析不同BMI患者两种方式配准误差的相关性及一致性。根据Van Herk公式计算两种配准方式的PTV外放边界。 结果 OSMS与CBCT组在腹背方向的误差差异有统计学意义（P < 0.05），而在左右、头脚平移方向及旋转RTN方向摆位误差差异无统计学意义（P > 0.05）。两种系统摆位误差尽管在统计学上显示有相关性（均P < 0.05），但仅在头脚方向和左右呈中度相关（r = 0.500、0.408），而在RTN旋转方向呈弱相关（r = 0.339），腹背方向呈极弱相关（r = 0.152）。两种方法一致性界限（LOA， 95% CI）在左右、头脚、腹背方向分别为［-0.45，0.45］、［-0.59，0.57］、［-0.48，0.40］ cm，在旋转RTN方向为-2.08° ~ 2.19°。不同BMI影响两种配准方式的结果，尤其在18.5 ~ 23.9 kg/m²的患者中，OSMS与CBCT在头脚方上呈强相关（r = 0.731）、左右方向呈中度相关（r = 0.512）、腹背及旋转RTN方向呈弱相关（r = 0.345、0.267），一致性检验显示两种影像系统摆位误差差异绝对值为0.4 ~ 0.5 cm/2°。采用CBCT及OSMS图像引导，在左右、头脚、腹背方向外放边界分别为［0.5 cm、0.7 cm、0.3 cm］和［0.5 cm、0.7 cm、0.5 cm］。 结论 在BMI为18.5 ~ 23.9 kg/m²的胸部肿瘤患者中，OSMS与CBCT两种图像引导方式在头脚及左右方向相关性较好，但一致性界限超过临床可接受范围；OSMS尚不能替代CBCT作为胸部肿瘤调强放疗的图像引导手段，进一步改进肿瘤运动替代物以增加OSMS图像引导的准确性是必要的。

Objective To evaluate the effect of optical body surface monitoring system （OSMS） in setup of intensity modulated radiotherapy （IMRT） for thoracic tumorsand to analyze its relationship with BMI. Methods Thesetup errors of CBCT and location CT with body membrane and the registration errors of OSMS and first body surface reference image without body membrane were obtained in 49 patients with thoracic tumor who received routine intensity modulated radiotherapy. The paired t-test was used to analyze the difference of registration errors between the two image guidance methods. Pearson′s correlation analysis was used to analyze the correlation between CBCT errors and OSMS errors， and Bland-Altman analysis was employed to evaluatethe agreement of the two errors. The correlation and consistency of the two registration errors in patients with different BMI index were analyzed. PTV external marginsby the two registration methods were calculated using Van Herk formula. Results The OSMS and CBCT groups demonstrated significant differences in setup errors in the ventrodorsal direction （P < 0.05）， while no significant differences were found in the left-right， head-foot translation directions， or RTN rotation directions （P > 0.05）. Although the two methods showed a significant correlation in setup errors （P < 0.05）， this correlation was only moderate in the head-foot and left-right directions （r = 0.500， 0.408）， weak in the RTN rotation direction （r = 0.339）， and very weak in the ventrodorsal direction （r = 0.152）. The limits of agreement （LOA， 95% CI） between the two methods were ［-0.45， 0.45］ cm in the left-right direction， ［-0.59， 0.57］ cm in the head-foot direction， and ［-0.48， 0.40］ cm in the ventrodorsal direction， with （-2.08° ~ 2.19°） in the RTN rotation direction.Different BMI levels influenced the results of the two registration methods， particularly in patients with a BMI of 18.5 ~ 23.9 kg/m2. In this group， OSMS and CBCT exhibited a strong correlation in the head-foot direction （r = 0.731）， a moderate correlation in the left-right direction （r = 0.512）， and weak correlations in the ventrodorsal and RTN rotation directions （r = 0.345， 0.267）. The absolute difference in setup errors between the two imaging systems was 0.4 ~ 0.5 cm/2°. Using CBCT and OSMS image guidance， the margins in the left-right， head-foot， and ventrodorsal directions were ［0.5 cm， 0.7 cm， 0.3 cm］ for CBCT， and ［0.5 cm， 0.7 cm， 0.5 cm］ for OSMS. Conclusion In chest tumor patients with a BMI of 18.5 ~ 23.9 kg/m2， OSMS and CBCT image guidance methods show good correlation in the head-foot and left-right directions， but their limits of agreement exceed the clinically acceptable range. OSMS cannot yet replace CBCT for image guidance in chest tumor intensity-modulated radiotherapy. Further improvements to tumor motion surrogates are necessary to enhance the accuracy of OSMS image guidance.