基于磁共振表观弥散系数分布的淋巴结良恶性鉴别方法研究

Study of differentiation of benign and malignant lymph nodes based on ADC value of MRI diffusion weighted imaging

目的以兔腘窝淋巴结动物模型,利用磁共振弥散加权成像,评价三种不同的半自动方法分析表观弥散系数(apparent diffusion coeffi cient,ADC)分布的临床价值。方法 21只兔被随机分为炎症和肿瘤转移两组,在动物模型建立以后,行磁共振扫描,采用多b值进行弥散加权成像,扫描后取出淋巴结行病理检查。利用线性回归计算表观弥散系数,比较良恶性淋巴结ADC值是否存在差异,并利用3种方式对图像进行分析:方法1:求淋巴结皮髓质ADC值之比;方法2:ADC值距人工标记中心距离的变化率;方法3:求ADC值随距形态学骨架距离的变化率。对其结果进行受试者操作特性曲线分析,计算相应的曲线下面积等统计参量。结果 41枚淋巴结被计入研究,22枚为转移,19枚为炎症。平均ADC值分别为:1.54×10-3mm2/s和1.42×10-3mm2/s(P=0.234)。曲线下面积三种方法及ADC平均值的曲线下面积分别为:0.839(皮髓质ADC值比),0.775(至人工中心),0.654(至骨架)和0.583(ADC平均值)。最大约登指数分别为0.639,0.517,0.304和0.266。结论皮髓质值比有最佳的曲线下面积,为三种方法中最佳,对评价淋巴结良恶性有一定临床价值。

Objective： To compare clinical value of three different semi-automatic methods in differential diagnosis of benign and malignant rabbit popliteal fossa nodes based on the apparent diffusion coefficient （ADC） map of magnetic resonance （MR） diffusion weighted imaging. Materials and Methods： Twenty-one rabbits were randomly divided into inflammatory and metastatic groups. After popliteal fossa lymph node metastasis model was setted up, MR diffusion weighted imaging was performed. After the scan the lymph nodes were taken for the pathologic diagnosis. The mean ADC of each whole lymph node was compared in the inflammatory and metastatic groups. Three different methods were used for evaluation of lymph nodes based on the ADC map. Method 1 ： The ADC ratio of cortex vs. medulla （high/ low） was calculated; Method 2： Curve of the mean ADC value vs. distance to the node＇s center, which was manually identified, was generated and slope of the curve was acquired. Method 3： Curve of the mean ADC value vs. distance to the node＇s morphological skeleton was generated and slope of the curve was acquired. Based on the pathological results, ROC was obtained and AUC was calculated in each procedure. Results： Forty-one lymph nodes were included in the experiment, 22 were metastatic and 19 were inflammatory. The mean ADC value of whole nodes in the two groups was 1.54×10^-3 mm^2/s and 1.42×10^-3 mm^2/s （P=0.234）. The area under the curve （AUC） in the three procedures and the mean of whole node were 0.839 （cortex vs. medulla）, 0.775 （to the center）, 0.654 （to the skeleton）, and 0.583 respectively. And the Youden index were 0.639, 0.517, 0.304, and 0.266. Conclusions： Of the 3 methods, Method 1 showed the best AUC and it might be the best semi-automatic ADC methods for the identification of lymph node malignancy.