摘要目的:利用形变配准技术(DIR)探究宫颈癌3种近距离放疗技术中，危及器官(OAR)高剂量区域的重叠性，比较DIR与直接叠加两种方法产生的OAR累积剂量差异，并分析该差异与重叠性的相关性。方法:回顾性选取45例接受近距离放疗的宫颈癌病例，分为腔内治疗组(ICBT)、腔内联合插植治疗组(IC/IS-BT)和3D打印导板组(3D)，每组15例。在计划系统中用布尔运算工具生成膀胱、直肠 D2 cm 3和 D1 cm 3对应的高剂量区域。对分次间的计划CT进行DIR，利用相似性系数(DSC)分析分次间膀胱、直肠高剂量区的重叠性。将所有分次计划CT形变配准在一起后，利用交并比(Jaccard)分析膀胱、直肠高剂量区的整体重叠性。DSC和Jaccard的取值范围为0~1，值越小表示重叠区域越小。利用DIR将所有分次计划剂量进行形变累积，与直接叠加法累积剂量进行比较，分析膀胱、直肠高量区的剂量差异与DSC、Jaccard的相关性。 结果:ICBT、IC/IS-BT和3D 3组膀胱 D2 cm 3、 D1 cm 3区域的DSC值分别为0.470±0.112、0.452±0.099、0.412±0.131和0.358±0.110、0.310±0.110、0.295±0.139，直肠 D2 cm 3、 D1 cm 3区域的DSC值分别为0.521±0.159、0.502±0.118、0.463±0.141和0.411±0.161、0.365±0.137、0.347±0.138；膀胱 D2 cm 3、 D1 cm 3区域的Jaccard值分别为0.110±0.056、0.076±0.059、0.083±0.067和0.049±0.039、0.032±0.031、0.042±0.028，直肠 D2 cm 3、 D1 cm 3区域的Jaccard值分别为0.167±0.130、0.114±0.089、0.110±0.094和0.107±0.095、0.057±0.065、0.056±0.067；3组技术的DSC、Jaccard值均较小，且差异均无统计学意义。经DIR后，膀胱和直肠 D2 cm 3和 D1 cm 3的累积剂量与直接叠加法累积剂量间的差异不明显，与重叠性没有相关性。 结论:3种治疗技术之间重叠性无明显差异，重叠性大小与剂量差异并不相关。形变叠加法可作为目前宫颈癌近距离放疗分次间总剂量评估的一项辅助方法。

abstractsObjective:To explore the overlap of high-dose regions in organs at risk (OARs) during the application of three brachytherapy techniques for cervical cancer using deformable image registration (DIR), to compare the differences in cumulative doses to OARs derived using DIR and the direct superposition method, and to analyze the correlation between these differences and the overlap.Methods:A retrospective analysis was conducted for 45 cervical cancer patients who received brachytherapy. Among them, 15, 15, and 15 cases were treated with intracavitary brachytherapy (the ICBT group), intracavitary/interstitial brachytherapy (the IC/IS-BT group), and 3D-printed applicator (the 3D group), respectively. The high-dose regions corresponding to the D2 cm 3 and D1 cm 3 in the bladder and rectum were generated using the Boolean operations in the planning system. DIR was performed for the interfractional planned CT images, followed by an analysis of the overlap of interfractional high-dose regions in the bladder and rectum using the dice similarity coefficient (DSC). After DIR was conducted for all fractional CT images, the overall overlap of high-dose regions in the bladder and rectum was analyzed using the intersection over union (also known as the Jaccard index). The values of both the DSC and Jaccard index varied from 0 to 1, with lower values representing smaller overlapping regions. Based on a comparison between the deformation-based cumulative doses of all fractional plans derived using DIR and the direct superposition method, the correlations of the dosimetric differences with the DSC and Jaccard index were analyzed for high-dose regions in the bladder and rectum. Results:The ICBT, IC/IS-BT, and 3D groups exhibited the DSC values of D2 cm 3 and D1 cm 3 of 0.470 ± 0.112, 0.452 ± 0.099, 0.412 ± 0.131 and 0.358 ± 0.110, 0.310 ± 0.110, 0.295 ± 0.139, respectively for the bladder, and 0.521 ± 0.159, 0.502 ± 0.118, 0.463 ± 0.141 and 0.411 ± 0.161, 0.365 ± 0.137, 0.347 ± 0.138, respectively for the rectum. These groups showed Jaccard indices of 0.110 ± 0.056, 0.076 ±0.059, 0.083 ± 0.067 and 0.049 ± 0.039, 0.032±0.031, 0.042±0.028, respectively for the bladder, and 0.167 ± 0.130, 0.114 ± 0.089, 0.110 ± 0.094 and 0.107 ± 0.095, 0.057 ± 0.065, 0.056 ± 0.067, respectively for the rectum. Therefore, the three groups showed both low DSC values and low Jaccard indices, with no statistically significant differences. Moreover, there were no significant difference between the cumulative doses to the bladder and rectum derived using DIR and the direct superposition method, demonstrating no correlation with the overlap of high-dose regions. Conclusions:There is no significant difference in the overlap of high-dose regions among the three brachytherapy techniques, with no correlation between the magnitude of the overlap and dosimetric differences. The deformation superposition method can serve as an auxiliary approach to assessing the total dose of all fractions in current brachytherapy for cervical cancer.