【摘要】 目的评估拉伸指数模型弥散加权成像(DWI)直方图分析术前预测单发肝细胞癌(HCC)微血管侵犯(MVI)的价值。方法回顾性分析43例经病理确诊单发HCC患者的腹部MRI,根据病理结果分为MVI(+)组(n=22)和MVI(-)组(n=21)。采用拉伸指数模型分析DWI,获得表观弥散系数(ADC)、分布弥散系数(DDC)和水分子各向异质性(α)直方图参数,比较组间临床资料及直方图参数差异;应用受试者工作特征(ROC)曲线分析差异有统计学意义的参数预测MVI的效能,并以多因素Logistic回归分析MVI的独立预测因素。结果组间肿瘤直径、ADC第5百分位数(5th)、DDC第5、25、50百分位数(5th、25th、50th)及平均值差异均有统计学意义(P均<0.05),其余直方图参数差异均无统计学意义(P均>0.05)。DDC 5th预测MVI的曲线下面积(AUC)最大,为0.86,高于DDC(25th、50th)及平均值、ADC 5th、肿瘤直径(Z=1.96～2.91,P均<0.05),其余参数两两比较差异均无统计学意义(Z=0.12～1.91,P均>0.05)。DDC 5th是MVI的独立预测因素(P=0.005)。结论拉伸指数DWI模型直方图分析有助于术前预测单发HCC的MVI,以DDC 5th诊断效能最佳。更多还原

【Abstract】 Objective To explore the value of histogram analysis of stretched exponential diffusion weighted imaging(DWI) model for predicting microvascular invasion(MVI) of single hepatocellular carcinoma(HCC) preoperation. Methods Abdominal MRI of 43 patients with single HCC diagnosed by pathology were retrospectively analyzed. The patients were divided into MVI(+) group(n=22) and MVI(-) group(n=21) according to pathologic findings. DWI was analyzed with stretched exponential model to obtain the histogram parameters, including apparent diffusion coefficient(ADC), distributed diffusion coefficient(DDC) and heterogeneity index(α). Clinical data and histogram parameters were compared between the groups. Receiver operating characteristic(ROC) curve of parameters being statistically different between the groups were drawn to evaluate the efficacies of predicting MVI, and multivariable logistic regression was used to analyze the independent predictors of MVI. Results Significant differences of HCC diameter, the 5 th percentile(5 th) of ADC, and the mean, 5 th, 25 th and 50 th percentiles(5 th, 25 th, 50 th) of DDC were found between groups(all P<0.05), while there was no statistical difference of other histogram parameters(all P>0.05). The area under the curve(AUC) of DDC 5 th for predicting MVI was 0.86, higher than that of the 25 th, 50 th and mean of DDC, ADC 5 th and HCC diameter(Z=1.96-2.91, all P<0.05), while there was no significant difference of AUC of other parameters(Z=0.12-1.91, all P>0.05). The 5 th of DDC was the independent predictor for MVI(P=0.005). Conclusion Histogram analysis of stretched exponential DWI model was helpful to preoperative prediction of MVI of single HCC, and the 5 th of DDC had the best diagnostic performance.更多还原