摘要
目的研究磁共振表观扩散系数(ADC)直方图在预测食管鳞癌病理分级(G)中的应用价值。方法回顾性分析术前行磁共振扩散加权成像(DWI)检查的食管鳞癌患者59例。入组患者32例,依据术后病理分级分为低级别组G1/2(16例)及高级别组G3(16例)。基于整个肿瘤体积进行ADC直方图分析,所得数据进行两组独立样本t检验和ROC曲线分析,计算直方图中ADC_(10)、ADC_(25)、ADC_(50)、ADC_(75)、ADC_(90)、ADC_(mean)、峰度、偏度和ROC曲线下面积(AUC)等值。结果低级别组中ADC_(10)=1.10±0.41×10^(-3)mm^2/s、ADC_(25)=1.37±0.47×10^(-3)mm^2/s,高级别组中ADC_(10)=1.49±0.33×10^(-3)mm^2/s、ADC_(25)=1.67±0.32×10^(-3)mm^2/s;高级别组中ADC10、ADC25高于低级别组且有统计学意义(P=0.006,0.044),ROC曲线分析示ADC_(10)与病理分级具有更好的相关性,能够预测高级别肿瘤(AUC 0.78;敏感度75.00%;特异度81.25%)。结论基于整个肿瘤体积的ADC直方图分析能够较好的解决肿瘤异质性问题,在一定程度上可以预测食管癌病理分级。
Objective To evaluate the value of apparent diffusion coefficient(ADC)histogram analysis in predicting the pathological grades of esophageal squamous carcinoma(ESC).Methods Preoperative diffusion-weighted MR imaging(DWI)of 59 patients with ESC was retrospectively analyzed.32 patients were included in this study and divided into 16 low-gradeⅠorⅡand 16 high-gradeⅢESC groups.ADC histogram parameters of the entire tumor volume including ADC10,ADC25,ADC50,ADC75,ADC90,ADCmean,kurtosis and skewness were determined.The parameters of the low-and high-grade groups were compared using two independent sample t-test and receiver operating characteristic(ROC)curves.Results The mean values of ADC10(1.49±0.33×10^-3 mm^2/s)and ADC25(1.67±0.32×10^-3 mm^2/s)in the high-grade group were significantly(P=0.006 and 0.044)higher than that in the low-grade group(1.10±0.41×10^-3 mm^2/s,1.37±0.47×10^-3 mm^2/s).ROC analyses showed that ADC10 was optimal for predicting high-grade ESC with area under the curve of 0.78,75.0%sensitivity and 81.25%specificity.Conclusion Histogram analysis of ADC maps of the entire tumor volume is useful in differentiating pathological grades of ESC.
作者
徐露露
孙娜娜
葛小林
甄福喜
孙新臣
XU Lu-lu;SUN Na-na;GE Xiao-lin;ZHEN Fu-xi;SUN Xin-chen(Nan Jing Medical University,Jiangsu 211166,China)
出处
《影像诊断与介入放射学》
2018年第6期440-444,共5页
Diagnostic Imaging & Interventional Radiology
基金
江苏省高校优势学科建设工程项目(JX10231801)
关键词
食管鳞癌
表观扩散系数
直方图分析
病理分级
Esophageal squamous carcinoma
Apparent diffusion coefficient
Histogram analysis
Pathological grade