【摘要】 目的:探讨基于MR图像的纹理分析技术在颅内纤维性肿瘤/血管周细胞瘤(SFT/HPC与血管瘤型脑膜瘤(AM)的术前鉴别诊断中的临床应用价值。方法:回顾性分析经手术及病理确诊的24例SFT/HPC患者和20例AM患者的临床资料及术前MRI资料。利用Image J软件在T2WI、对比增强T1WI(T1CE)和ADC图像中选取肿瘤最大径层面并在瘤体实性部分勾画ROI,采用直方图和灰度共生矩阵(GLCM)两种纹理分析方法,测量病灶的最大值、最小值、均值、中位数、标准差、峰度、偏度、角二阶矩(ASM)、对比度、逆差矩、熵值及自相关等纹理参数,并采用两独立样本t检验和MannWhitney U检验比较SFT/HPC与AM的各项纹理参数的差异,对有统计学意义的参数绘制受试者工作特征(ROC)曲线,分析其对SFT/HPC及AM的鉴别诊断效能。结果:①直方图参数中,在T2WI、T1CE和ADC图像上SFT/HPC的最大值(MaxT2WI、MaxT1CE)、最小值(MinT2WI、MinT1CE、MinADC)、均值(MeanT2WI、MeanT1CE、MeanADC)和中位数(MedT2WI、MedT1CE、MedADC)均低于AM,在T1CE图像上的标准差(SDT1CE)大于AM,在T2WI和T1CE图像上的峰度值(KurtT2WI、KurtT1CE)和偏度值(SkeT2WI、SkeT1CE)均大于AM,组间差异均有统计学意义(P<0.05)。②在GLCM参数中,SFT/HPC在ADC图像上的ASM值(ASMADC)小于AM,在T2WI和T1CE图像上的对比度(ConT2WI、ConT1CE)小于AM,在ADC图像上的熵值(EntADC)大于AM,组间差异均有统计学意义(P<0.05);在T2WI、T1CE和ADC图像上的逆差矩值(IDMT2WI、IDMT1CE、IDMADC)及T1CE图像上的自相关值(CorT1CE)的组间差异均有统计学意义(P<0.05);③在AUC>0.7的参数中,以MinT1CE、MedT1CE和MeanT1CE的诊断效能最佳,AUC(敏感度、特异度)分别为0.970(95.8%、90.0%)、0.964(91.7%、95.0%)和0.960(91.7%、95.0%)。结论:MRI的纹理分析可提供更多量化信息,对于术前通过MRI鉴别SFT/HPC与AM有着重要的临床意义。更多还原

【Abstract】 Objective:To explore the clinical value of preoperative MRI image texture analysis in distinguishing solitary fibrous tumors/hemangiopericytoma(SFT/HPC)and angiomatous meningioma(AM).Methods:The clinical data and preoperative magnetic resonance imaging data of 32 patients with SFT/HPC and 24 patients with AM confirmed by operation and pathology were retrospectively analyzed.All patients underwent T2 WI,DWI and contrast-enhanced T1 WI(T1 CE).The texture parameters,including maximum value,minimum value,average value,median,standard deviation(SD),skewness,kurtosis,angular second moment,contrast,inverse different moment,entropy and correlation were measured using histograms and grey-level co-occurrence matrix(GLCM)and compared between SFT/HPCs and AMs.The independent sample t test or the Mann-Whitney Utest was used for data procession and analysis.ROC curves for statistically significant parameters were used to analyse their efficacy in predicting SFT/HPC and AM.Results:①In histogram parameters,the maximum values(MaxT2 WI,MaxT1 CE),minimum values(MinT2 WI,MinT1 CE,MinADC),mean values(MeanT2 WI,MeanT1 CE,MeanADC),and medians(MedT2 WI,MedT1 CE,MedADC)of SFT/HPCs measured on T2 WI,T1 CE,and ADC images were smaller than those of AM;and the kurtosis values(KurtT2 WI,KurtT1 CE)and skewness values(SkeT2 WI,SkeT1 CE)of SFT/HPCs measured on T2 WI and T1 CE were all greater than those of AMs,and the difference was statistically significant(P<0.05).②In GLCM parameters,the ASM energy value(ASMADC)measured on ADC images,and the contrast values(ConT2 WI,ConT1 CE)measured on T2 WI and T1 CE of SFT/HPCs were smaller than those of AM;and the entropy values(EntADC)of SFT/HPCs on ADC images were greater than that of AMs,and the difference was statistically significant.The inverse moment values on T2 WI,T1 CE and ADC images(IDMT2 WI,IDMT1 CE,IDMADC)and the autocorrelation values on T1 CE(CorT1 CE)were statistically different between the two groups(all P<0.05);③For the texture parameters with statistical difference between the two groups,ROC curves were drawn,and in the parameters with the area under the curve(AUC)>0.7,the AUC(sensitivity and specificity)of MinT1 CE,MedT1 CEand MeanT1 CEwas 0.970(95.8%,90.0%),0.964(91.7%,95.0%)and 0.960(91.7%,95.0%),respectively.Conclusion:This study demonstrated that MRI texture analysis can provide more quantitative information,which has important clinical significance for distinguishing SFT/HPCs and AMs.更多还原