IVIM-DWI鉴别诊断肝细胞癌和肝局灶性结节增生

Evaluation of intravoxel incoherent motion DWI in differential diagnosis of hepatocellular carcinoma and focal nodular hyperplasia

目的探讨体素内不相干运动扩散加权成像(IVIM-DWI)鉴别肝细胞癌(HCC)与肝局灶性结节增生(FNH)的价值。方法对407例临床疑似HCC或FNH的患者行常规上腹部MR平扫、动态增强及IVIM-DWI扫描,60例患者(40例HCC,20例FNH)入组。IVIM采用单、双指数模型获得表观扩散系数(ADC)、慢速表观扩散系数(D)、快速表观扩散系数(D*)及快速扩散成分所占比例(f)。结果FNH组的ADC、D、D*及f值分别为(1.60±0.25)×10^(-3) mm^2/s、(1.12±0.17)×10^(-3) mm^2/s、(44.89±18.23)×10^(-3) mm^2/s和(34.80±9.68)%;HCC组分别为(1.32±0.21)×10^(-3) mm^2/s、(0.82±0.21)×10^(-3) mm^2/s、(49.82±20.11)×10^(-3) mm^2/s和(28.72±13.84)%。2组间的ADC、D值差异有统计学意义(P均<0.001),而D*、f值差异无统计学意义(P>0.05)。D值相应ROC曲线下面积为0.90,以0.96×10^(-3) mm^2/s为阈值诊断HCC的敏感度、特异度分别为84.44%、90.02%。结论 IVIM-DWI有助于鉴别诊断HCC和FNH,其中双指数模型计算的D值诊断效能更高。

Objective To explore the feasibility of intravoxel incoherent motion DWI （IVIM-DWI） in differential diagnosis of hepatocellular carcinoma （HCC） and focal nodular hyperplasia （FNH）. Methods A total of 407 patients with clinically-suspected HCC or FNH underwent conventional and dynamic enhanced MRI and IVIM-DWI, 60 patients （40 cases of HCC, 20 cases of FNH） were enrolled. Parameters of ADC, slow apparent diffusion coefficient （D）, fast apparent diffusion coefficient （D ） and fraction of fast apparent diffusion coefficient （f） were obtained by monoexponential model and biexponential model respectively. Re- suits Thevalues of ADC, D, D＂ and f in FNH group were （1.60±0.25） × 10^-3mm2/s, （1.12±0.17）× 10^-3mm2/s, （44.89±18.23）× 10^-3 mm2/s and （34.80 ±9.68）%, and those in HCC group were （1.32 ± 0.21） × 10^-3 mm2/s, （0.82±0. 21）× 10^-3mm2/s, （49.82±20. 11）× 10^-3mm2/s and （28. 72±13.84）%, respectively. Significant inter-group differ- ences were observed in ADC and D （both P〈0. 001）, however, there were no significant differences in Ds and f （both P〈0. 05）. The areas under the ROC curve of D were 0. 90, and taking D： 0. 96 × 10^- 3 mm2/s as cut-off value, the sensitivity and specificity of D in diagnosis of HCC were 84. 44% and 90. 02%. Conclusion IVIM-DWI is useful to distinguish FNH from HCC, and the D value in biexponential model has the best diagnostic efficacy for differentiations.