磁共振定量参数在鉴别Ⅰ型与Ⅱ型子宫内膜癌中的价值

Value of quantitative MRI parameters in differentiating type Ⅰ from type Ⅱ endometrial carcinoma

目的:综合运用Logistic回归分析及ROC曲线方法,探讨磁共振定量参数在鉴别Ⅰ型与Ⅱ型子宫内膜癌中的价值。方法:回顾性分析2015年4月—2020年3月经保定市第一中心医院病理证实为子宫内膜癌的87例患者的术前磁共振图像及临床资料。由2名影像科主治医师分别测量Ⅰ型与Ⅱ型子宫内膜癌的平均表观弥散系数(apparent diffusion coefficient,ADC),最大径(the longest diameter,LD),比较Ⅰ、Ⅱ型子宫内膜癌的ADC及LD的差异。利用Pearson等级相关分析ADC及LD与Ⅰ、Ⅱ型子宫内膜癌的相关性,并运用Logistic回归联合ROC曲线评估ADC及LD对不同分型子宫内膜癌的鉴别效能。结果:Ⅰ型68例(78.2%),Ⅱ型19例(21.8%),不同分型子宫内膜癌ADC及LD的差异有统计学意义(P<0.05)。ADC与组织学分型呈负相关(r=-0.528,P<0.001),LD与组织学分型呈正相关(r=0.254,P=0.018)。Logistic回归分析结果显示ADC可独立预测不同分型子宫内膜癌(P<0.001),LD (P=0.449)不能独立预测。当ADC≤0.974×10^(-3)mm^(2)/s时的AUC为0.887,灵敏度为76.5%、特异度为94.7%。结论:ADC对Ⅰ、Ⅱ型子宫内膜癌的鉴别有一定的价值,综合利用Logistic回归分析及ROC曲线能够提高鉴别Ⅰ、Ⅱ型子宫内膜癌的准确性。

Objective To determine the feasibility of quantitative MRI parameters in differentiating type Ⅰ from typeⅡ endometrial carcinoma by model of Logistic Regression and receiver operating characteristic(ROC) curve.Methods In this study,the MR images of 87 patients with a confirmed diagnosis of endometrial carcinoma,on the basis of surgical pathology results,treated at Baoding First Central Hospital from April 2015 to March 2020 were analyzed retrospectively.According to tumor type,the ADC and the longest diameter(LD) of tumor was compared.The correlation between different types and various parameters were evaluated with Pearson’s correlation analysis.The efficacy of various parameters in differentiating type Ⅰ from type Ⅱ endometrial carcinoma was evaluated by Logistic regression analyses and ROC curve.Results The different type was type Ⅰ in 68 cases(78.2%),type Ⅱ in 19 cases(21.8%).The difference of ADC and LD of type Ⅰ,type Ⅱ were statistically significant(P=0.018 <0.001).There was a negative correlation between ADC and tumor type(r=-0.528,P <0.001),and a positive correlation between LD and tumor type(r=0.254,P=0.018).Logistic regression analysis showed that ADC independently predicted type I from type II endometrial carcinoma(P<0.001),LD can’t(P=0.449).When ADC was less than 0.974×10^(-3) mm^(2)/s,the area under the ROC curve was 0.887,with a sensitivity of 76.5%,specificity of 94.7%.Conclusion The quantitative MRI parameters ADC is useful for differentiating type Ⅰ from type Ⅱ endometrial carcinoma.Applications of logistic regression and ROC curve increase diagnostic accuracy in differentiating the pathological type of endometrial carcinoma before surgery.