利用T_1加权DCE-MRI技术对脑胶质瘤分级

Study of Glioma's Grades Using T_1-weighted DCE-MRI Technique

应用T1加权动态对比度增强磁共振成像(DCE-MRI)技术对脑胶质瘤进行术前分级,验证其用于脑胶质瘤分级的准确性和可重复性。对26例脑胶质瘤患者团注Gd-DTPA对比剂后,以1.5TSiemens Syngo MRI扫描仪采集DCE T1加权图像。首先基于改进的Tofts两室药物动力学模型,从采集到的动态组图像建立病灶区感兴趣区内的平均信号强度-时间曲线,然后应用非线性最小二乘拟合方法对该曲线进行拟合,获取定量参数。这些参数包括:初始增强率ER、回流速率常数Kep、排泄速率常数Kel、曲线下面积AUC、峰值高度PH、到达峰值的时间TTP,最后与临床组织病理学结果进行比较,建立这些定量参数与脑胶质瘤分级之间的定量关系。采用独立样本t检验比较任意两个分级之间各定量参数的统计学差异,P<0.05作为统计学差异标准。ER、AUC和PH随着脑胶质瘤的病理分级由低到高地增加;低级别脑胶质瘤中的Ⅰ与Ⅱ级之间以及高级别中的Ⅲ与Ⅳ级之间的比较结果显示,ER、AUC和PH的值均无明显统计学差异(P>0.05);低级别与高级别级之间的比较结果显示,仅参数ER的差异具有统计学意义(P<0.05)。ER是术前准确界定脑胶质瘤低级别与高级别的最具敏感性和特异性的参数,其界定阈值约为0.68。

This project is purposed to evaluate the grade of a glioma by employing T1-weighted dynamic enhanced magnetic resonance imaging （DCE-MRI） technique before surgery, and to assess the accuracy and reproducibility of the study of applying T1- weighted DCE-MRI to differential the grades. 26 patients with gliomas were injected by Gd-DTPA and imaged by applying a 1.5 T Siemens Syngo MRI scanner and T1- weighted DCE-MRI method. All the DCE T1-weighted data acquired were processed via pharmacokinetic analysis based on modified Tofts＇ two compartment model and nonlinear least square fitting method, from which signal-time curves of region of interest （ROI） selected in region of focal intensifying were drawn and some quantitative parameters such as initiale enhancement rate （ER ）, the flux rate constant from extravascular extracellular space to plasma space （ Kop ） , the elimination rate （ Kel ） , the area under the curve （ AUC ） , peak height （PH） and time to peak （TTP） correlating vessel endothelium permeability were measured. An independent-samples t-test for these parameters was performed by using SPSSll. 5 statistic software. A P less than 0.05 for a parameter was considered as statistically significant. En , AUC or PH increased lineally with the glioma grading from Ⅰ to Ⅳ ; in comparing Ⅰ and Ⅱ of the low grade or Ⅲ and Ⅳ of the high grade, ER, AUC and PH had no statistically significant difference （P 〉 0. 05）. However, only ER was statistical significant （P 〈 0.05 ） compared with the low grade with the high grade. ER was the most sensitivity and specificity parameter to accurately distinguish the low with the high grade, and the threshold of the quantitative parameter for distinguishing the low with the high grade was found as 0.68.