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

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 coefficient,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...

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

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.