摘要目的:评估统计过程控制（Statistical process control，SPC）技术在宫颈癌放疗患者分次间误差管理中的可行性。方法:纳入2024年4月至2025年3月在本中心治疗的宫颈癌放疗患者，图像引导频率为前5次加后续每周1次。回顾性地对所有患者分次间三个方向的平移误差制作单值（Individual，I）、移动极差（Moving Range，MR）及指数加权移动平均（Exponential Weighted Moving Average，EWMA）控制图。并以所有患者误差数据的每日均值制作三个方向上整个周期时间排序的单值-移动极差（I-MR）控制图，依据控制图分析误差数据的波动，变异性及偏移趋势。采用Minitab 22软件进行统计分析，使用±5 mm作为规格界限进行能力分析，分别计算未触发无行动阈值（No Action Level，NAL）协议患者的误差数据（Base组）、触发NAL协议患者的前五次误差数据（Pre组）及后四次误差数据（Post组）三个方向的整体能力度量 Pnpk。 结果:研究共纳入112例宫颈癌患者的误差数据。I、MR及EWMA控制图中存在超出控制限外的异常分次的患者分别占总患者的10.71%，10.71%和15.18%，其中第9次CBCT扫描分次的异常分次在总异常分次中占比为52.83%。时间排序的I-MR控制图显示，X和Z方向的异常点主要集中于整个周期的第1-24 d及第208 d后，Y方向的上在整个周期前期移动极差控制图异常点有集中现象。Base组在左右（Lat，X）、头脚（Lng，Y）及前后（Vrt，Z）三个方向的 Pnpk值分别为1.01、0.71和0.90；Pre组为0.61、0.34和0.86；Post组为1.27、0.73和0.83。 结论:宫颈癌患者疗程后期误差数据会存在一定的系统性偏移风险。在基于±5 mm的过程能力分析中，以±3 mm为阈值触发的NAL协议的系统误差修正干预措施可有效提升X和Y方向上放射治疗中的系统误差控制能力，可适当增加位置验证频次以提高治疗精度。结合SPC方法中的控制图及能力分析对摆位误差进行过程监测与评估，能够为图像引导放疗（image guided radiation therapy，IGRT）中患者摆位准确性与稳定性控制提供量化支持，具有良好的临床应用价值。

abstractsObjective:Evaluate the feasibility of Statistical Process Control (SPC) techniques in managing interfraction errors in cervical cancer radiation therapy.Methods:Patients with radiotherapy for cervical cancer treated in our center from April 2024 to March 2025 were included. The image-guided frequency was the first five times plus the follow-up once a week. Retrospectively create Individual (I), Moving Range (MR), and Exponential Weighted Moving Average (EWMA) control charts for the translational errors in three directions between each fraction for all patients. A single value moving range (I-MR) control chart for the whole cycle time sequence in three directions was made based on the daily mean of all patients' error data, and the fluctuation, variability and offset trend of error data were analyzed according to the control chart. The data were statistically analyzed by Minitab 22 software. The tolerance of ± 5 mm was used as the specification limit for capability analysis, and the overall capability Pnpk measures in three directions were calculated: the error data of patients who did not trigger the no action level (NAL) protocol (Base group), the first five error data of patients who triggered the NAL protocol (Pre group) and the last four error data (Post Group). Results:A total of 112 patients with cervical cancer were included in the study. In the I, MR, and EWMA control charts, the proportions of patients with abnormal fractions beyond the control limits were 10.71%, 10.71%, and 15.18%, respectively. Among all abnormal fractions, the 9th CBCT scan fraction accounted for 52.83%. The time ordered I-MR control chart shows that the abnormal points in the X and Z directions are mainly concentrated in the 1th to 24th days and 208 days of the whole cycle, and the abnormal points in the Y direction are concentrated in the early stage of the whole cycle. In the Base group, the Pnpk values in the left and right (LAT, X), head and foot (LNG, Y) and front and rear (VRT, Z) directions were 1.01, 0.71 and 0.90, respectively; 0.61, 0.34 and 0. 86 in the Pre group; The Post group was 1.27, 0.73 and 0.83. Conclusions:There is a certain risk of systematic bias in the error data of cervical cancer patients in the later stage of treatment. In the process capability analysis with control limits set at ±5 mm, the NAL protocol triggered by a ±3 mm threshold for systematic error correction-demonstrated an improvement in process capability for controlling systematic setup errors in the X and Y translational directions during radiotherapy. The frequency of position verification may be appropriately increased to improve treatment accuracy. Combined with the control chart and ability analysis of SPC method, the process monitoring and evaluation of positioning error can provide quantitative support for the accuracy and stability control of patients' positioning in image guided radiation therapy (IGRT), which has good clinical application value.