摘要
目的探讨多b值扩散加权序列的三指数模型拟合相关参数3D兴趣区直方图分析在胶质瘤术前分级中的价值。方法分析2016-08—2018-02经病理证实的46例(低级别组17例,高级别组29例)胶质瘤患者的MRI影像资料,分别测量三维ROI直方图分析三指数模型相关参数,对低级别组和高级别组上述参数进行比较。结果低级别和高级别组Ds-P5、Ds-P10、Ff-mean和Ff-median比较差异均有统计学意义,分别为:Ds-P5(0.76±0.08)×10^(-3) mm^2/s vs (0.73,0.06)×10^(-3) mm^2/s(IQR,P=0.010);Ds-P10(0.84±0.10)×10^(-3) mm^2/s vs (0.80,0.09)×10^(-3) mm^2/s (IQR,P=0.013);Ff-mean 0.20±0.04 vs 0.22±0.04(P=0.033);Ff-median 0.15±0.06 vs 0.20±0.04(P=0.003)。结论三指数模型3D兴趣区直方图分析相关参数可以很好地进行胶质瘤术前分级。
Objective To explore the value of three-dimension histogram method for tri exponential model related parameters in preoperative grading of glioma.Methods From Aug.2016 to Feb.2018,46 patients(28 men,18 women;10-72 years old;17 low grade gliomas,29 high grade gliomas)received a pathologic diagnosis and MR scan.The three-dimension histogram method of tri-exponential model related parameters based on multi b value DWI sequence were measured,and then compared the difference between low and high grade gliomas.Results Ds P5,Ds P10,Ff mean and Ff median showed obvious difference between low and high grade gliomas,Ds P5 value were(0.76±0.08)×10^-3 mm^2/s and(0.73,0.06)×10^-3 mm^2/s(IQR,P=0.010),Ds P10 value were(0.84±0.10)×10^-3 mm^2/s and(0.80,0.09)×10^-3 mm^2/s(IQR,P=0.013),Ff mean value were 0.20±0.04 and 0.22±0.04(P=0.033),Ff median value were 0.15±0.06 and 0.20±0.04(P=0.003),respectively;but other parameters showed no differences.Conclusion Three dimension histogram method for tri-exponentioal related parameters of Ds value and Ff value might be of value to differentiate low and high grade gliomas.
作者
李建红
杜勇
肖翔
张静
许乙凯
王春红
LI Jianhong;DU Yong;XIAO Xiang;ZHANG Jing;XU Yikai;WANG Chunhong(Department of Medical Imaging,Xinyang Central Hospital,Xinyang 464000,China;Department of Medical Imaging,Nanfang Hospital,Southern Medical University,Guangzhou 510515,China)
出处
《中国实用神经疾病杂志》
2019年第1期34-39,共6页
Chinese Journal of Practical Nervous Diseases
关键词
胶质瘤
扩散加权成像
三指数模型
ROC曲线
直方图
Glioma
Diffusion weighted imaging
Tri-exponential model
Receiver operating characteristic curve
Histogram