常规CT特征联合直方图分析预测急性肾损伤的价值

Value of conventional CT features combined with histogram analysis in predicting acute kidney injury

目的:探讨CT平扫特征及直方图分析诊断梗阻性急性肾损伤(AKI)的价值。方法:收集我院急性单侧尿路结石梗阻患者77例,其中急性肾损伤组40例,非急性肾损伤组37例。应用Siemens区域增长功能,测量常规CT平扫特征(包括梗阻侧、非梗阻侧肾脏体积比,结石体积、位置,肾周炎)和CT平扫直方图参数(包括Kurtosis、Skewness、Variance、Mean、1%、10%、50%、90%、99%百分位值)。计数资料采用χ^(2)检验,计量资料采用t检验(符合正态分布)或Mann-Whitney U检验(符合偏态分布)进行组间比较,将组间差异显著的参数纳入二元logistic回归分析,评估预测AKI的独立预测因素。采用受试者操作特征(ROC)曲线评价各参数价值;采用Spearman相关分析,比较各参数与估算肾小球滤过率(eGFR)的相关性。结果:AKI与非AKI组比较,常规CT特征(结石体积、肾周炎、肾脏体积比)及直方图参数(Kurtosis、Skewness、Variance及90%百分位值)具有统计学意义(P均<0.05);二元Logistic回归分析显示,肾脏体积比、Kurtosis、Variance是预测AKI的独立因素;ROC曲线显示联合参数诊断AKI效能较高,ROC曲线下面积(AUC)为0.94,特异度89.2%,敏感度82.5%,高于单一参数模型。相关性分析显示肾脏体积比、Kurtosis、Skewness、Variance、90%百分位值与eGFR相关性较高。结论:常规CT联合直方图分析预测AKI的效能较高,可用于无创性评估急性肾损伤。

Objective:To explore the value of conventional CT features combined with histogram analysis in predicting acute kidney injury.Methods:CT of 77 patients with acute unilateral urinary tract stone obstruction were reviewed.There were 40 in AKI group and 37 in non-AKI group.The regional growth function of Siemens post-processing technology was used to measure the characteristics of conventional CT plain scan(including the volume ratio of kidney on obstructive side and non-obstructive side,stone volume,location and perinephritis)and CT plain scan histogram(including kurtosis,skewness,variance,mean value,and 1%,10%,50%,90%and 99%percentile values).Counting data were compared byχ^(2) test.The measurement data were compared between groups by t-test(normal distribution)or Mann Whitney U-test(skew distribution).The parameters with significant differences between groups were included in binary logistic regression analysis to determine the independent predictors of AKI.The receiver operating characteristic(ROC)curve was used to evaluate the diagnostic performance.Pearson and Spearman analysis was used to determine correlation between imaging parameters and estimated glomerular filtration rate(eGFR).Results:Compared with non AKI group,conventional CT parameters(stone volume,perinephritis,kidney volume ratio)and histogram parameters(kurtosis,skewness,variance and 90%percentile value)were statistically significant in AKI and non-AKI groups(P<0.05).Logistic regression analysis showed that renal volume ratio,kurtosis and variance were independent factors for AKI prediction.ROC curve showed higher performance of combined parameters in diagnosis of AKI,with AUC of 0.94,specificity of 89.2%and sensitivity of 82.5%,respectively.eGFR showed high correlation with thickening bridging septa in the perinephric space,renal volume ratio,kurtosis,skewness,variance and 90%percentile value were highly correlated with.Conclusion:Conventional CT combined with histogram analysis is effective tool in predicting AKI.