感兴趣区设置方法的不同对肾透明细胞癌CT值测量的影响

Tumor CT Measurements in Clear Cell Renal Cell Carcinoma:Influence of Different ROI Methods on Tumor Attenuation

目的评估感兴趣区（ROI）设置方法的不同对肾透明细胞癌患者CT值测量及其一致性的影响。方法回顾性分析经病理证实为肾透明细胞癌的50例患者资料。观察者在轴面平扫、动脉期、静脉期及延迟期CT图像上采用3种不同ROI设置方法（肿瘤轮廓法、最大层面法、肿瘤多个小样本法）对肾透明细胞癌病灶行CT值测量,3个月后该观察者以同样方法重复病灶各期CT值的测量。采用单因素方差分析比较3种不同ROI勾画方法所测各期CT值的差异,然后对不同ROI设置方法测量的平均CT±SD值及观察者两次测量的重复性及一致性用Bland-Altman法及组内相关系数（ICC）法进行对比分析。结果观察者采用3种不同ROI设置方法所测各期CT值均存在较好的一致性（ICC〉0.7）,且三者之间的差异均有统计学意义（P〈0.05）。Bland-Altman散点图显示肿瘤轮廓法测定的各期CT值分布均更集中,提示肿瘤轮廓法在同一观察者重复测量中一致性区间小,准确性好。结论肾透明细胞癌患者采用肿瘤轮廓法与肿瘤最大层面法及多个小样本法勾画选择ROI所得CT值有显著差异,采用肿瘤轮廓法重复测量一致性评估差异度小,可重复性好。

Objective To assess the influence of different region of interest （ROI） methods on tumor attenuation and the measurement consistency in patients with clear cell renal cell carcinoma. Methods Fifty cases of clear cell renal cell carcinoma patients confirmed by pathology were collected. The CT attenuation was measured according to three ROI proto- cols ： freehand outline ROI, single-slice or small solid samples by the same researcher for two times. The three protocols were compared for differences in CT attenuation and the interclass correlation coefficient （ICC） were also calculated.. Intraob- server variability using ANOVA and Bland-Ahman plot. Results The three methods of ROI set had significant differences from each other The CT attenuation for small solid samples were significantly higher than the other two ROIso The CT attenuation obtained by ROIs for freehand outline ROI, single-slice and small solid sample had good consistency（ ICC 〉 0.7 ）. ICC for the freehand outline ROI was excellent（ ICC 〉 0.9）. The results of Bland-Ahman plot show that methods of freehand outline ROI is the most reproducible . Conclusion There were statistically significant differences among the three different ROI methods. The tumor analysis by outlined ROI protocol can provide more reproducible and stable results.