表观扩散系数直方图分析鉴别间变型脑膜瘤和非典型脑膜瘤

Apparent diffusion coefficient histogram analysis in differentiating anaplastic meningioma from atypical meningioma

目的:探讨表观扩散系数(ADC)直方图分析对间变型脑膜瘤(Anaplastic meningioma,AnM)和非典型脑膜瘤(Atypical meningioma,AtM)的鉴别诊断价值。方法:回顾性收集经组织病理证实的AtM(n=42)和AnM(n=24)患者的临床、影像和病理资料。利用MaZda软件在轴位ADC图像上对整个肿瘤进行勾画并自动生成以下直方图参数,包括平均值(ADC_(mean))、变异度(AD-Cvariance)、偏度(ADC_(skewness))、峰度(ADC_(kurtosis))、第1百分位数(ADC_(p1))、第10百分位数(ADC_(p10))、第50百分位数(ADC_(p50))、第90百分位数(ADC_(p90))和第99百分位数(ADC_(p99))。通过绘制受试者工作特征(Receiver operating characteristic,ROC)曲线来评估其在术前鉴别AnM与AtM的诊断效能。结果:AnM的ADCp1、ADC_(p10)、ADC_(p50)均小于AtM(P均<0.05),ADC_(mean)、ADC_(variance)、ADC_(skewness)、AD-Ckurtosis、ADC_(p90)和ADC_(p99)在两组间差异无统计学意义(P均>0.05)。ROC曲线分析显示,当ADCp1以75×10^(-3) mm^(2)/s作为截断值时,鉴别AnM和AtM的AUC、敏感性、特异性、正确率、阳性预测值和阴性预测值分别为0.768(0.648~0.863)、58.3%、92.9%、80.3%、82.4%和79.6%。结论:ADC直方图分析有助于术前无创鉴别AnM和AtM,其中以ADCp1的诊断效能最高。

Objective:To evaluate the value of apparent diffusion coefficient(ADC)histogram analysis in the differential diagnosis of anaplastic meningioma(AnM)and atypical meningioma(AtM).Methods:The clinical,imaging and pathological data of AtM(n=42)and AnM(n=24)were collected retrospectively.Using MaZda software,the whole tumor was sketched on the axial ADC image and the following histogram parameters were generated automatically,including ADC_(mean),ADC_(variance),ADC_(skewness),ADC_(kurtosis),1st percentile(ADC_(p1)),10th percentile(ADC_(p10)),50th percentile(ADC_(p50)),90th percentile(ADC_(p90)),99th percentile(ADC_(p99)).The receiver operating characteristic(ROC)curve was drawn to evaluate its diagnostic efficacy in differentiating AnM from AtM preoperatively.Results:ADC_(P1),ADC_(P10),ADC_(P50)of AnM were all lower than those of AtM(P0.05).There was no significant difference in the degree of ADC_(mean),ADC_(variance),ADC_(skewness),ADC_(kurtosis),ADC_(p90),ADC_(p99)between the two groups(P0.05).ROC curve analysis showed that when AD_(Cp1)was cut off at 75×10^(-3)mm^(2)/s,the AUC,sensitivity,specificity,accuracy,positive predictive value and negative predictive value of differentiation between AnM and AtM were 0.768(0.648~0.863),58.3%,92.9%,80.3%,82.4%and 79.6%,respectively.Conclusion:ADC histogram analysis is helpful to differentiate AnM from AtM,and ADC_(P1)is the most effective.