IVIM双指数、拉伸指数模型对子宫内膜癌分型的价值

目的 探讨体素内不相干运动（IVIM）双指数、拉伸指数模型各参数在Ⅰ型与Ⅱ型子宫内膜癌（EC）中的鉴别诊断价值。 方法 回顾性分析51例经手术病理证实的Ⅰ型（31例）与Ⅱ型（20例）EC病人的IVIM成像资料。由2名医师在IVIM各参数伪彩图上分别测量慢速表观扩散系数（ADC-slow）、快速表观扩散系数（ADC-fast）、灌注分数（f）、扩散分布指数（DDC）和扩散异质性指数（α）。采用组内相关系数（ICC）评价2名医师测量参数的一致性。采用Fisher确切概率法比较Ⅰ型与Ⅱ型EC的病理类型及分级分期。绘制受试者操作特征（ROC）曲线,分析2组间有统计学差异的参数及其诊断效能。 结果 2名医师所测数据的一致性较好（均ICC≥0.75）。Ⅰ型组的ADC-slow、DDC值均高于Ⅱ型组,ADC-fast值低于Ⅱ型组（均P<0.05）。ADC-slow值为0.500×10^-3 mm²/s时,鉴别Ⅰ型和Ⅱ型EC的AUC为0.926,敏感度、特异度、准确度分别为80.0%、90.3%、86.3%。 结论 双指数模型的参数ADC-slow、ADC-fast值和拉伸指数模型的参数DDC值均有助于鉴别Ⅰ型与Ⅱ型EC,其中ADC-slow值鉴别诊断效能更大。

Value of intravoxel incoherent motion MR imaging with biexponential and stretched-exponential models in differentiating types of endometrial cancer

Objective To explore the intravoxel incoherent motion (IVIM) MR imaging with biexponential and stretched-exponential models in differentiating type Ⅰ and type Ⅱ of endometrial cancer. Methods IVIM MR imaging data of 31 patients with endometrial cancer type Ⅰ and 20 patients with type Ⅱ confirmed by pathologically were analyzed retrospectively. IVIM parameters including the slow apparent diffusion coefficient (ADC-slow), fast apparent diffusion coefficient (ADC-fast), perfusion fraction (f), distributed diffusion coefficient (DDC), and heterogeneity of intravoxel diffusion (α) on pseudo-color map were measured by two radiologists. The intra-class correlation coefficient(ICC) was performed to assess the consistency of their measurements. The Fisher’s exact test were used to compare the differences in pathological stage and classification between type Ⅰ and type Ⅱ. The receiver operator characteristic (ROC) curve was plotted to identify the parameters which were significant in differentiating type Ⅰ from Ⅱ, and their diagnostic efficacies. Results Excellent agreement was found between two radiologists (both ICC≥0.75). The ADC-slow and DDC were higher and the ADC-fast was lower in the type Ⅰ than in the type Ⅱ (both P<0.05). The AUC of ADC-slow for differentiating type Ⅰ from Ⅱ was 0.926, with optimal threshold of 0.500×10^-3 mm²/s, it achieved a sensitivity 80.0%, specificity 90.3% and accuracy 86.3%. Conclusion The ADC-slow value and ADC-fast value in the biexponential model and the DDC value in stretched-exponential model can be used for differentiating type Ⅰ and Ⅱ of endometrial cancer, and the ADC-slow value performs best.