RRIMS:乳腺X线筛查中的辐射风险——模型评估
评估RRIMS模型预测的乳腺X线筛查腺体平均剂量与手工计算剂量的一致性,纳入317名女性共733次筛查。Bland-Altman分析显示偏差有统计学意义(均值偏差-2.02%,一致性界限40.24%至-44.27%),ICC(3,1)为0.64(P<0.001),Kendall's W为0.83(P<0.001),结论为两者一致性有统计学意义且尚可。
要点
- RRIMS预测剂量与手工计算剂量的ICC(3,1)=0.64,Kendall's W=0.83,一致性尚可。
- Bland-Altman显示偏差均值-2.02%,一致性界限上限40.24%、下限-44.27%。
- 本研究仅针对Hologic一家厂商设备,多厂商适用性有待未来版本研究。
核心结论
- 手工计算剂量与RRIMS预测剂量之间的一致性有统计学意义且尚可。
- 本研究仅针对Hologic一家主要乳腺X线设备厂商,模型未来版本有潜力开展多厂商适用性研究。
研究设计与样本
- 研究类型为模型评估,比较RRIMS模型预测的腺体平均剂量与手工计算剂量。
- 样本为317名女性,共733次筛查访视,覆盖1至4轮筛查。
- 采用定性与定量统计分析评估两种剂量值的百分比差异及一致性,定性用Bland-Altman图,定量用ICC(3,1)与Kendall's W。
诊断标准与关键数值
- Bland-Altman分析显示手工计算与RRIMS预测剂量百分比差异的偏差均值(bias)为-2.02%,一致性界限上限40.24%、下限-44.27%,偏差有统计学意义。
- ICC(3,1)为0.64(P<0.001),Kendall's W为0.83(P<0.001)。
局限性
- 本研究仅针对Hologic一家主要乳腺X线设备厂商,多厂商适用性有待未来版本研究。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
英文原文(摘要)· 点击展开
OBJECTIVE: The Radiation Risk In Mammography Screening (RRIMS) model was introduced as a novel tool to help females accurately calculate their lifetime mean glandular dose (MGD) and estimate their population-level risk of radiation-induced breast cancer incidence and mortality.
METHODS: The model's accuracy was evaluated by comparing the received MGD of 317 women who had undergone a total of 733 visits across one to four rounds of screening. This was achieved by comparing the RRIMS predicted dose values with the same examination dose calculated manually by hand. Qualitative and quantitative statistical analyses were performed to assess the percentage difference (% diff) or agreement between the two values.
RESULTS: Qualitative statistical analysis using the Bland-Altman plots demonstrated a statistically significant bias for the % diff between the manually calculated and RRIMS predicted dose values, where the mean (bias) was -2.02% with an upper and lower limit of agreement of 40.24% and -44.27%, respectively. Quantitative statistical analysis revealed an intraclass correlation coefficient (ICC, 3,1) of 0.64 (p-value < 0.001) and a Kendall's W of 0.83 (p-value < 0.001).
CONCLUSION: The results indicate a statistically significant and reasonably good level of agreement between the manually calculated vs RRIMS predicted dose values. This work was focused on one of the major mammography equipment manufacturers that is Hologic, however there is potential for a multivendor applicability study of this model with future iterations. This will further improve upon this innovative dose and risk prediction tool that can empower healthcare professionals when making informed decisions and enhance patient care.
ADVANCES IN KNOWLEDGE: This paper assesses the precision of the dose and risk model that our team has previously established. The results bring us one step closer to providing females and clinicians with a useful tool that can help explain and contextualise the benefits and risks associated with screening mammography.
引用
- Hooshmand S, Reed WM, Suleiman ME, et al. RRIMS: Radiation Risk In Mammography Screening - model evaluation. Br J Radiol. 2023. PMID: 37750941. doi:10.1259/bjr.20230250
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。