基于常规MRI图像的纹理分析鉴别:血管周细胞瘤/孤立性纤维瘤与血管瘤型脑膜瘤

Texture analysis of conventional MRI parameters for differentiating between hemangioma meningioma and hemangiopericytoma based on whole tumor measurement

目的基于肿瘤全域对比分析颅内血管瘤型脑膜瘤(hemangioma meningioma,HM)与血管周细胞瘤/孤立性纤维瘤(hemanyiopericytoma,HPC)的T1WI、T2WI以及T1增强图像纹理参数的差异,并评估纹理分析方法对两者的鉴别诊断价值。材料与方法回顾性分析经手术及病理证实的8例HPC患者与9例HM患者的临床资料及术前磁共振成像资料。所有患者均行T1WI、T2WI平扫以及T1增强扫描。利用Omni-Kinetics软件后处理后得到3D感兴趣区的T1、T2以及T1增强信号强度直方图信息及其所有纹理参数,进行组间比较,并利用受试者操作特性(receiver operating characteristic,ROC)曲线评价各参数的鉴别诊断效能。结果在T1增强信号强度图像纹理参数中的最小值、平均值、中位数、均方根值、平均差、偏度、均一性、第5、10、25、75、90百分位数,以及T2信号强度图像纹理参数中的偏度、均一性、第5、10、25百分位数、集群阴影、集群突出在两者间差异有统计学意义(P<0.05)。利用ROC曲线分析,T2信号中以均一性0.79为阈值鉴别二者诊断能力最佳(曲线下面积=1.00),敏感性及特异性分别为88.9%、100%。结论基于肿瘤全域感兴趣区T2以及T1增强信号强度图像的纹理分析可以为血管周细胞瘤与血管瘤型脑膜瘤之间的鉴别提供更多信息。

Objective:To comparatively analyze the differences of T1WI,T2WI and contrasted T1WI signal intensity histogram parameter in the differential diagnosis of intracranial hemangioma meningioma(HM)and hemanyiopericytoma(HPC)based on entire tumor,and further to evaluate the differential diagnosis value between them.Materials and Methods:The conventional MRI data of 8 patients with HM and 9 patients with HPC proven by pathology or clinical follow-up were retrospectively reviewed.All patients underwent T1WI,T2WI and contrasted T1WI scan.The texture features of lesions based on axial T1WI,T2WI and contrasted T1WI were extracted by post-processed 3D ROI with Omni-Kinetics software.The texture parameters were compared using two independent samples t test and Mann-Whitney U test between groups.Receiver operating characteristic curve(ROC)was constructed to assess the differential ability of the significant parameters between HM and HPC.Results:Min intensity,mean value,mean deviation,skewness,median intensity,RMS,uniformity,the 5th,10th,25th,75th and 90th percentiles of contrasted T1WI have significant differences between HM and HPC(P<0.05).Uniformity,skewness,the 5th,10th,25th,cluster shade and cluster prominence of T2WI have significant differences between the two(P<0.05).According to ROC analysis,the uniformity of T2WI was the best parameters(cutoff value=0.79,AUC=1.0)for distinguishing HM and HPC of 0.79 with sensitivity of 88.9%and specificity of 100%.Conclusions:Texture analysis of T2WI and contrasted T1WI based on entire tumor should be helpful in differentiation of HM and HPC.