不同类型全容积ADC值评估宫颈癌病理分级的效能分析

Efficacy Analysis of Different Types of Full-volume ADC Values to Evaluate the Pathological Grading of Cervical Cancer

目的比较不同类型的全容积ADC值评估宫颈癌病理分级的效能。方法回顾分析58例宫颈癌患者术前DWI资料,包括3例高分化、37例中分化、18例低分化,在GE Adw4.6工作站沿肿瘤边缘手动勾画ROI获得肿瘤所有层面的ADC值,取其平均值作为肿瘤的平均ADC值。计算标准化ADC值(nADC),nADC=肿瘤平均ADC值/参照物(臀大肌)ADC值。利用Omni-kinetics纹理分析软件获得全容积ADC直方图参数。比较高/中分化组和低分化组宫颈癌各类型ADC值差异,有统计学差异的参数绘制ROC曲线,比较曲线下面积(AUC)。结果低分化组的平均ADC值、nADC值、ADCmin、ADCmean、各百分位数ADC值均小于高/中分化组,差异有统计学意义(Mann-Whitney U检验,P分别=0.025、0.026、0.000、0.006、0.003~0.032)。两组间ADCmax值没有统计学差异(独立样本t检验,P=0.529)。ADCmin有最佳的鉴别诊断效能,AUC为0.836,其后依次为5^(th)ADC值,AUC 0.757;10^(th)ADC值,AUC 0.748;平均ADC值,AUC 0.745;25^(th)ADC值,AUC 0.730;nADC值,AUC 0.726;50^(th)ADC值、75^(th)ADC值,AUC均为0.723;90^(th)ADC值,AUC 0.713;95^(th)ADC值,AUC 0.688。ADCmin以0.584×10^(-3)mm^(2)/s作为最佳截断值,灵敏度、特异度分别为94.6%、62.5%。结论ADC值测量有助于评估宫颈癌病理分级,ADCmin有最佳的鉴别诊断效能。

Objective To analyze the efficacy of different types of full-volume ADC values in differentiating the pathological grading of cervical cancer.Methods Preoperative DWI data of 58 patients with cervical cancer were retrospectively analyzed,including 3 patients with well differentiation,37 patients with moderate differentiation,and 18 patients with poor differentiation.ROI was manually plotted along the edge of the tumor at GE Adw4.6 workstation to obtain ADC values of all layers of the tumor,and the average value of all layers was taken as the average ADC value of the tumor.Normalized ADC(nADC),nADC=tumor average ADC value/reference(gluteus maximus)ADC value were calculated.Omni-kinetics texture analysis software was used to obtain ADC full-volume histogram parameters.The values of various types of ADC in the well/moderately differentiated group and the poorly differentiated group were compared.ROC curves were drawn for the parameters with statistical differences,and the area under the curve(AUC)was calculated.Results The mean ADC value,nADC value,ADCmin,ADCmean,and ADC value of each percentage in the poor differentiation group were all lower than those in the well/moderate differentiation group,with statistically significant differences(mann-whitney U test,P=0.025,0.026,0.000,0.006,0.003-0.032,respectively).There was no statistical difference in ADCmax between the two groups(independent sample t test,P=0.529).The ADCmin had the best differential diagnosis efficiency,with the AUC of 0.836,followed by the ADC value of 5th,and the AUC of 0.757;10^(th)ADC value,AUC 0.748;average ADC value,AUC 0.745;25^(th)ADC value,AUC 0.730;nADC value,AUC 0.726;50^(th)ADC value and 75^(th)ADC value,AUC 0.723;90^(th)ADC value,AUC 0.713;95^(th)ADC value,AUC 0.688.ADCmin uses 0.584×10^(-3) mm^(2)/s as the best cutoff value,and the sensitivity and specificity are 94.6%and 62.5%,respectively.Conclusion ADC value measurement is helpful to evaluate the pathological grade of cervical cancer,and ADCmin has the best differential efficacy.