头颅CT扫描中心层面水当量直径估算体型特异性辐射剂量的可行性

Validity of the size-specific dose estimate derived from the water equivalent diameter value of the center slice of the scan range in the head CT examination

目的探讨中心层面水当量直径（Dw）估算体型特异性辐射剂量（SSDE）在头颅CT检查中的可行性。 方法回顾性分析2017年3月至4月浙江省人民医院临床拟诊头颅疾患需行头颅CT平扫的197例患者。测量所有层面Dw，标定转换系数（f），记录每例患者的容积CT剂量指数（CTDIvol）值，同时对每个层面管电流进行校准，并计算所有层面水当量直径（Dw-gro）、中心层面水当量直径（Dw-cen）及相对应的SSDE（SSDEgro和SSDEcen）。以散点图表示扫描Z轴方向Dw、f、CTDIvol、SSDE的变化。采用Pearson相关分析及线性回归分析评价Dw-gro与Dw-cen的相关性；采用Spearman相关分析及线性回归分析评价SSDEgro与SSDEcen、SSDE与Dw、CTDIvol与Dw的相关性。以SSDEgro作为参考值，计算SSDEcen估算辐射剂量的平均绝对差（MARD）评估准确性，并采用多元逐步线性回归分析探讨MARD相关影响因素。 结果近颅顶层面的体型最小，对应最大的f及SSDE值，标称CTDIvol值最小。Dw-gro与Dw-cen、SSDEgro与SSDEcen、SSDE与Dw、CTDIvol与Dw均呈正相关，r值分别为0.947、0.931、0.416、0.626，P均〈0.05。Dw-gro与Dw-cen分别为（16.94±0.69）、（18.50±0.62）cm，SSDEgro与SSDEcen分别为54.10（52.29，56.39）、53.77（51.85，55.25）mGy。SSDEcen估算剂量的总体MARD为1.62%。以MARD值为应变量，f、Dw、CTDIvol为自变量行多元逐步线性回归分析，f未能进入回归方程，MARD值与Dw呈独立负相关（β=–1.319，P〈0.05），与CTDIvol呈独立正相关（β=0.202，P〈0.05）。 结论以Dw-cen计算的SSDEcen具有低MARD值，适用于头颅CT扫描患者的辐射剂量的估算。

ObjectiveTo explore the validity of the size-specific dose estimate （SSDE） derived from the water-equivalent diameter （Dw） value of the slice located in the middle of the scan range in the head CT examination. MethodsA total of 197 patients underwent head CT nonenhanced scan were enrolled in this retrospective study. The Dw, size-dependent conversion factor （f）, normalized volume CT dose index （CTDIvol） and SSDE values of all slices were calculated. Two sets of SSDE, SSDEgro and SSDEcen based on the Dw values slice by slice （Dw-gro） and the Dw values of the slices in the middle of the scan range （Dw-cen）, were obtained across all patients. Pearson correlation analysis and linear regression analysis were performed for Dw-gro vs Dw-cen, Spearman correlation analysis and linear regression analysis for SSDEgro vs SSDEcen, SSDE vs Dw, CTDIvol vs Dw. With the reference of SSDEgro value, mean absolute relative difference （MARD） of SSDEcen values were calculated to assess its accuracy and the correlated factors of MARD was analyzed with multivariate linear stepwise regression analysis. ResultsThe minimal Dw value close to the roof of the skull corresponded to the maximal value of f and SSDE, which was the minimal value of CTDIvol. The significant positive correlation was showed between Dw-gro vs Dw-cen,SSDEgro vs SSDEcen,SSDE vs Dw,CTDIvol vs Dw （r=0.947, 0.931, 0.416, 0.626;P〈0.05）. The values of Dw,gro and Dw-cen were （16.94±0.69） and （18.50±0.62） cm respectively. The values of SSDEgro and SSDEcen were [54.10 （52.29, 56.39）] mGy and [53.77 （51.85, 55.25）] mGy respectively. An approximation of SSDEcen values with an average of 1.62% of the gross MARD was found to match the reference value. Multivariate linear stepwise regression analysis indicated that MARD had negative correlation with Dw （β=– 1.319,P〈0.05） , positive correlation with CTDIvol （β=0.202,P〈0.05） , and f was not included in the multivariate regression equation. ConclusionSSDEcen based on the Dw value of the slice located at the center of the scan range yields small MARD value and can represent a reliable SSDE estimation in the head CT examination.