【作者】 邹启桂；

【导师】 徐海波；

【作者基本信息】 华中科技大学 ， 影像医学与核医学， 2011， 博士

【摘要】 第一部分3.0T高分辨率磁敏感加权成像评估幕上脑胶质瘤磁敏感效应：与磁共振灌注成像(PWI)对比研究目的：探讨3.0T高分辨率SWI磁敏感效应程度和PWI-rCBV对幕上脑胶质瘤分级的准确性及二者的相关性。材料与方法：经病理证实的44例幕上脑胶质瘤患者(29例低级别,15例高级别)行常规MR平扫和增强检查、非对比增强磁敏感加权成像(Susceptibility weighted imaging,SWI)和首过动态磁敏感对比增强磁共振灌注成像(Dynamic susceptibility contrast perfusion weighted imaging,DSC-PWI)。在肿瘤实质灌注最明显的区域及对侧正常脑组织区域测量局部组织的的脑血容量(CBV)；根据幕上脑胶质瘤的磁敏感效应程度将其分为不同的级别。评估胶质瘤的磁敏感效应程度与肿瘤分级的相关性及准确性；评估rCBV与脑胶质瘤分级的关系；根据ROC曲线选取合适临界值,评估rCBV对脑胶质瘤分级的准确性；评估脑胶质瘤磁敏感效应程度与rCBV的相关性。所有数据采用SPSS13.0软件包处理,统计前对样本数据进行正态分布及方差齐性检验。两样本均数比较采用独立样本t检验,相关性分析采用Pearson积差相关分析和Spearman等级相关分析,配对资料采用精确概率法检验(Fishers exact test), P<0.05认为差异具有统计学意义。结果：1.低、高级别脑胶质瘤肿瘤实质的磁敏感效应程度存在明显统计学差异(P<0.01)；高级别胶质瘤肿瘤实质rCBV为(4.15+1.21),明显高于低级别胶质瘤(2.62+1.59)(P<0.01)；低、高级别少突胶质细胞瘤肿瘤实质的磁敏感效应程度存在统计学差异(<0.05)而rCBV在二者之间无明显统计学差异(>0.05)。2.低、高级别星形细胞瘤肿瘤实质的磁敏感效应程度、rCBV均存在明显统计学差异(P<0.01)；二者与组织病理学的相关性分别为:r=0.723, P<0.01; r=0.645, P<0.01.3.星形细胞瘤肿瘤实质磁敏感效应程度与rCBV的相关性为：r=0.594,P<0.01。4.根据磁敏感效应进行星形细胞瘤分级,其灵敏度为90.9%,特异度为90.5%,PPV为83.3%,NPV为95%,错分类百分比为9.4%,(假阳性率+假阴性率)为18.6%；ROC曲线分析,其曲线下面积为0.916,P<0.01。当rCBV取2.38时,其分级灵敏度为100%,特异度为71.4%,PPV为64.7%,NPV为100%,错分类百分比为18.8%,(假阳性率+假阴性率)为28.6%；ROC曲线分析,其曲线下面积为0.892,P<0.01。5.分别用SWI方法和PWI方法进行星形细胞瘤分级的结果与常规MR方法分级结果比较,两种方法均存在明显统计学差异(P<0.01)；用SWI方法的分级结果与PWI分级结果比较,两种方法间亦存在明显统计学差异(P<0.01)。结论：星形细胞瘤肿瘤实质的磁敏感效应程度与rCBV有明显的相关性；运用SWI方法和PWI-rCBV方法均可以提高星形细胞瘤分级的准确性,而磁敏感效应程度分级诊断效能更高。对少突胶质细胞瘤,SWI或许可以用于其分级。第二部分3.0T高场多种常用磁共振功能成像方法对胶质瘤分级的临床应用价值初探第一节首过动态磁敏感对比增强磁共振灌注成像对幕上脑胶质瘤分级的临床应用价值初探目的：探讨3.0T动态磁敏感对比增强磁共振灌注成像(DSC-PWI)对幕上脑胶质瘤分级的临床应用价值。材料与方法：经病理证实的41例幕上脑胶质瘤(低级别28例,高级别13例)行常规MR平扫+增强检查,DSC-PWI检查。分别测量肿瘤实质及对侧正常脑实质的CBV，CBF,计算出肿瘤实质的rCBV (rCBV=肿瘤实质CBV/对侧正常脑组织CBV)、rCBF (rCBF=肿瘤实质CBF/对侧正常脑组织CBF)；根据肿瘤实质灌注曲线计算信号强度恢复百分比。根据ROC曲线选取rCBV、rCBF、信号强度恢复百分比最合适分级临界值,分别评估肿瘤实质rCBV、rCBF、信号强度恢复百分比与胶质瘤病理分级的相关性和对胶质瘤分级的准确性。所有数据采用SPSS13.0软件包处理,统计前对样本数据进行正态分布及方差齐性检验。两样本均数比较采用独立样本t检验,相关性分析采用Spearman等级相关分析,配对资料采用精确概率法检验(Fisher’s exact test),P<0.05认为差异具有统计学意义。结果：1.高级别胶质瘤肿瘤实质的rCBV (4.29±1.23)明显高于低级别胶质瘤(2.71±1.58)(P<0.01)；高级别胶质瘤肿瘤实质信号强度恢复百分比(0.56±0.17)明显低于低级别胶质瘤(0.78±0.12)(P<0.01)。低、高级别少突胶质细胞瘤肿瘤实质的rCBV、rCBF、信号强度恢复百分比均无明显统计学差异(P>0.05)。低、高级别星形细胞瘤肿瘤实质的rCBV (2.22±0.92和4.01±1.13)、信号强度恢复百分比(0.80±0.05和0.54±0.19)均存在明显统计学差异,rCBF在两者之间无统计学差异。2.星形细胞瘤肿瘤实质rCBV和信号强度恢复百分比与组织病理学的相关性分别为：r=0.633, P<0.01;r=0.686, P<0.01。ROC曲线下面积分别为：0.894,P<0.01；0.928,P<0.01。3.星形细胞瘤肿瘤实质rCBV临界值取2.35、信号强度恢复百分比取0.689时,其分级结果与常规方法均存在明显差异(P<0.05)。信号强度恢复百分比分级结果明显优于常规MRI方法。结论：星形细胞瘤肿瘤实质rCBV、信号强度恢复百分比可以用于肿瘤分级,可以提高常规MRI分级的准确性,信号强度恢复百分比这一指标对肿瘤分级效能更高。第二节1H MRS对幕上脑胶质瘤分级的临床应用价值初探目的：探讨3.0T氢质子磁共振波谱成像(1HMRS)对幕上脑胶质瘤分级的临床应用价值。材料与方法：经病理证实的43例幕上脑胶质瘤(低级别30例,高级别13例)行常规MR平扫+增强检查和1H MRS检查。分别测量肿瘤实质及对侧正常脑实质的NAA/Cr,、Cho/Cr、Cho/NAA。根据ROC曲线选取NAA/Cr,、Cho/Cr、Cho/NAA最合适分级临界值,分别评估肿瘤实质NAA/Cr,、Cho/Cr、Cho/NAA与胶质瘤病理分级的相关性和对胶质瘤分级的准确性。所有数据采用SPSS13.0软件包处理,统计前对样本数据进行正态分布及方差齐性检验。两样本均数比较采用独立样本t检验,相关性分析采用Spearman等级相关分析,配对资料采用精确概率法检验(Fishers exact test), P<0.05认为差异具有统计学意义。结果：1.高级别胶质瘤肿瘤实质的NAA/Cr (0.52±0.27)、Cho/NAA (5.47±3.35)与低级别胶质瘤肿瘤实质NAA/Cr (1.05±0.59)、Cho/NAA (2.63±1.61)存在明显统计学差异(P<0.05)。NAA/Cr,、Cho/Cr、Cho/NAA在低、高级别少突胶质细胞瘤肿瘤实质间均不存在统计学差异(P>0.05)。低、高级别星形细胞瘤肿瘤实质的NAA/Cr (1.14±0.63、0.58±0.28)、Cho/NAA(2.02±1.06、4.95±3.31)均存在明显统计学差异(P<0.01)2. NAA/Cr,、Cho/NAA与星形细胞瘤组织病理学的相关性分别为：r=0.563,P<0.01；r=0.635,P<0.01。其分级ROC曲线下面积分别为：0.85,P<0.01；0.895,P<0.01。3.当星形细胞瘤肿瘤实质NAA/Cr取值0.51,Cho/NAA取值2.76时,其分级结果与常规方法分级结果比较,均无明显差异；但Cho/NAA分级的灵敏度、特异度、阳性预测值、阴性预测值明显高于常规方法和NAA/Cr方法,错分类百分比及(假阳性率+假阴性率)明显低于常规方法和NAA/Cr方法。Cho/NAA和NAA/Cr两种方法分级结果比较存在明显统计学差异。结论：NAA/Cr,、Cho/NAA均可用于幕上星形细胞瘤分级,其NAA/Cr分级准确性同常规方法相当,Cho/NAA可以进一步提高分级准确性。第三节ADC值对幕上脑胶质瘤分级的临床应用价值初探目的：探讨3.0T磁共振弥散加权成像的ADC值对幕上脑胶质瘤分级的临床应用价值。材料与方法：经病理证实的35例幕上脑胶质瘤(低级别23例,高级别12例)行常规MR平扫+增强检查和DWI检查。在ADC图上分别测量肿瘤实质及对侧正常脑实质的ADC值。根据ROC曲线选取ADC值最合适分级临界值,分别评估肿瘤实质ADC值与胶质瘤病理分级的相关性和对胶质瘤分级的准确性。所有数据采用SPSS13.0软件包处理,统计前对样本数据进行正态分布及方差齐性检验。两样本均数比较采用独立样本t检验,相关性分析采用Spearman等级相关分析,配对资料采用精确概率法检验(Fisher s exact test), P<0.05认为差异具有统计学意义。结果：1.低、高级别胶质瘤肿瘤实质的ADC值明显高于对侧正常脑实质；高级别胶质瘤肿瘤实质ADC值((1134.05+135.59)×10-6mm2/s)明显低于低级别胶质瘤肿瘤实质((1472.93±326.52)×10-6mm2/s) (P<0.01)。低、高级别少突胶质细胞瘤肿瘤实质ADC值无统计学差异(P>0.05)。低、高级别星形细胞瘤肿瘤实质ADC值((1580.56±308.77)×10-6mm2/s)和(1130.73±10.20)×10-6mm2/s) )存在明显统计学差异(P<0.01)2.星形细胞瘤肿瘤实质ADC值与组织病理学分级的相关性为：r=0.695,P<0.01。其分级ROC曲线下面积为0.922,P<0.01。3.当星形细胞瘤肿瘤实质ADC取值(1241.05)×10-6mm2/s时,其分级结果与常规方法分级结果比较,存在明显统计学差异；其分级的灵敏度、特异度、阳性预测值、阴性预测值明显高于常规方法,而错分类百分比及(假阳性率+假阴性率)明显低于常规方法。结论：ADC值可用于幕上星形细胞瘤分级,其分级准确性高于常规MR方法。第三部分3.0T MRI幕上脑胶质瘤分级优化指标的选择目的：探讨3.0T磁共振灌注成像(PWI)、磁敏感加权成像(SWI)、氢质子波谱成像(1H MRS)、弥散加权成像对幕上脑胶质瘤分级的最优化指标的选择。材料与方法：经病理证实的32例幕上脑胶质瘤(低级别21例,高级别11例)行常规MR平扫+增强检查、PWI、SWI、MRS和DWI检查。根据幕上脑胶质瘤的磁敏感效应程度(SED)将其分为不同的级别。分别测量肿瘤实质及对侧正常脑实质的Cho/NAA、ADC值,根据肿瘤实质灌注曲线计算出信号强度恢复百分比。根据ROC曲线选取Cho/NAA、ADC值及信号强度恢复百分比最合适分级临界值,分别评估肿瘤实质SED、Cho/NAA、ADC值及信号强度恢复百分比与胶质瘤病理分级的相关性和对胶质瘤分级的准确性。所有数据采用SPSS13.0软件包处理,统计前对样本数据进行正态分布及方差齐性检验。两样本均数比较采用独立样本t检验,相关性分析采用Pearson积差相关分析和Spearman等级相关分析,配对资料采用精确概率法检验(Fishers exact test), P<0.05认为差异具有统计学意义。结果：1.低、高级别胶质瘤肿瘤实质的Cho/NAA、ADC值、信号强度恢复百分比及SED均存在明显统计学差异(P<0.05)。2.低、高级别少突胶质细胞瘤肿瘤实质SED存在明显差异(P<0.05),余三个指标均无明显统计学差异。3.低、高级别星形细胞瘤肿瘤实质上述四个指标均存在明显统计学差异(P<0.01)。星形细胞瘤肿瘤实质Cho/NAA、ADC值、信号强度恢复百分比及SED与组织病理学分级的相关性为：r=0.695, P<0.01; r=0.757, P<0.01; r=0.757, P<0.01;r=0.815, P<0.01。4.应用星形细胞瘤肿瘤实质Cho/NAA、ADC值、信号强度恢复百分比、磁敏感效应程度进行胶质瘤分级,其中ADC值、信号强度恢复百分比及SED三个指标其分级的灵敏度、特异度、阳性预测值、阴性预测值均高于常规方法、错分类百分比及(假阳性率+假阴性率)均低于常规方法。SED和信号恢复百分比两个指标的分级结果亦存在明显统计学差异(P<0.01)。结论：星形细胞瘤肿瘤实质SED、Cho/NAA、ADC值及信号强度恢复百分比均可以提高胶质瘤分级的准确性；SED和信号强度恢复百分比对星形细胞瘤分级具有最高的诊断效能,而SED具有最高的临床应用价值。更多还原

【Abstract】 PartiAssessment of susceptibility effect using 3.0T High Field High Resolution Susceptibility Weighted Imaging in Patients with supratentorial Gliomas:Comparison with MR Perfusion ImagingObjective:Our aim was to determine whether the degree of susceptibility effect (SED) on high-resolution susceptibility-weighted imaging (HR-SWI) could be used for suparatentorial glioma grading, assess the correlates with maximum relative cerebral blood volume (rCBVmax) and to compare its diagnostic accuracy for glioma grading with that of dynamic susceptibility-weighted contrast-enhanced perfusion MR imaging (DSC).Methods and materials:44 suparatentorial gliomas (29 for low-grade,15 for high-grade) confirmed by craniotomy surgery and histopathology underwent conventional MRI plus enhancement, non contrast enhanced HR-SWI and DSC-PWI. Measured the maximum relative cerebral blood volume (rCBV) in tumor parenchyma with the most high perfusion region. According to the SED classified the suparatentorial glioma in four grades. Assessed the correlation of SED, rCBV with pathologic grade of suparatentorial glioma respectively and compared their diagnostic accuracy for glioma grading. Assessed the correlation of SED and maximum rCBV in parenchyma of tumor. All the data were used SPSS 13.0 for statistics, tested the Homogeneity of variance and Normal distribution before statistcs. We used Independent samples t test, Pearson and Spearman correlation analysis and Fishers exact test, and P<0.05 was considered Statistically significant.Results:1. SED of tumor parenchyma between low-grade and high-grade gliomas had significant difference(P<0.01); rCBV in parenchyma of high-grade glioma was 4.15±1.21,which was obvious higher than that of low-grade glioma (2.62±1.59) (P<0.01). SED of tumor parenchyma between low-grade and high-grade Oligodendroglial tumours had significant difference(P<0.05) but rCBV had no significant difference(P>0.05).2. SED and rCBV of tumor parenchyma between low-grade and high-grade astrocytoma had significant difference respectively (P<0.01); both of them had significant correlation with pathology grade (P<0.01, respectively).3. The SED and rCBV had significant correlation in parenchyma of astrocytoma (P<0.01).4. Using SED for astrocytoma grade, the sensitivity was 90.0%, specificity was 90.5%, positive predictive value (PPV) was 83.8%, negative predictive value (NPV) was 95%, fraction of misclassified (FM) was 9.4%,and (false positive ratio plus false negative ratio) was 18.6%,the area of Roc Curve was 0.916;when rCBV was 2.38, the grading sensitivity was 100%, specificity was 71.4%, PPV was 64.7%, NPV was 100%, fraction of misclassified (FM) was 18.8%,and (false positive ratio plus false negative ratio) was 28.6%, the area of Roc Curve was 0.892.5. There were significant difference among SWI method, PWI-rCBV method and conventional MRI method for grade astrocytoma and SWI method was the best one.Conclusion:The SED and rCBV had significant correlation in parenchyma of astrocytoma; both SWI method and PWI-rCBV method could improve diagnostic accuracy for grading and SED was better than PWI-rCBV. For Oligodendroglial tumours, SED may used for grading. Part 2 In the assessment of supratentorial glioma grade by using 3.0T high-field multiplex function MRI methodsThe first segment In the assessment of supratentorial glioma grade by using 3.0T high field dynamic susceptibility weighted contrast-enhanced perfusion MR imagingObjective:Our aim was to determine whether the dynamic susceptibility-weighted contrast-enhanced perfusion MR imaging (DSC) could be used for suparatentorial glioma grading and assess its diagnostic accuracy.Methods and materials:41 suparatentorial glioma (28 for low-grade,13 for high-grade) confirmed by craniotomy surgery and histopathology underwent conventional MRI plus enhancement and DSC-PWI. Measured the maximum relative cerebral blood volume (rCBV), maximim relative cerebral blood flow (rCBF) in tumor parenchyma with the most high perfusion region. Calculated the percentage of signal intensity recovery derived from perfusion curve. Selected the most appropriate threshold values according to ROC curves. Assessed the correlation of rCB V, rCBF and the percentage of signal intensity recovery with pathologic grade of suparatentorial glioma respectively and compared their diagnostic accuracy for glioma grading. All the data were used SPSS 13.0 for statistics, tested the Homogeneity of variance and Normal distribution before statistics. We used Independent samples t test, Spearman correlation analysis and Fisher’s exact test, and P<0.05 was considered Statistically significant.Results:1. rCBV in tumor parenchyma of high-grade glioma(4.29±1.23) was obvious higher than that of low-grade glioma (2.71±1.58) (P<0.01), and the percentage of signal intensity recovery in high-grade glioma (0.56±0.17) was significant lower than that of low-grade (0.78±0.12) (P<0.01). rCBV, rCBF and the percentage of signal intensity recovery in tumor parenchyma of high-grade Oligodendroglial tumours had no difference compared with low-grade. rCBV and the percentage of signal intensity recovery in tumor parenchyma of high-grade astrocytoma had significant difference with that of in low-grade.2. rCBV and the percentage of signal intensity recovery in tumor parenchyma of astrocytoma had significant correlation with pathology grade (P<0.01,respectively).3. when rCBV was 2.35 and the percentage of signal intensity recovery was 0.689, both of their grading results had significant difference compared with conventional MRI method, and the grading result according to the percentage of signal intensity recovery was obvious better than conventional MRI method.Conclusion:rCBV and the percentage of signal intensity recovery in tumor parenchyma of astrocytoma could be used for grading, which could improve the diagnostic accuracy for grading than conventional MRI, and the percentage of signal intensity recovery has the higher efficacy. The second segment In the assessment of supratentorial glioma grade by using 3.0T high-field Multi-voxel proton MR spectroscopyObjective:Our aim was to determine whether the Multi-voxel proton MR spectroscopy (1H MRS) could be used for suparatentorial glioma grading and assess its diagnostic accuracy.Methods and materials:43 suparatentorial gliomas (30 for low-grade,13 for high-grade) confirmed by craniotomy surgery and histopathology underwent conventional MRI plus enhancement and 1H MRS. Measured the NAA/Cr, Cho/Cr, Cho/NAA in tumor parenchyma and their Contralateral normal brain tissue. Selected the most appropriate threshold values according to ROC curves. Assessed the correlation of NAA/Cr,、Cho/Cr、Cho/NAA with pathologic grade of suparatentorial glioma respectively and compared their diagnostic accuracy for glioma grading. All the data were used SPSS 13.0 for statistics, tested the Homogeneity of variance and Normal distribution before statistics. We used Independent samples t test, Spearman correlation analysis and Fisher’s exact test, and P<0.05 was considered Statistically significant.Results:1. There were significant difference of NAA/Cr and Cho/NAA in tumor parenchyma between high-grade and low-grade glioma(P<0.05). NAA/Cr, Cho/Cr and Cho/NAA in tumor parenchyma of high-grade Oligodendroglial tumours had no difference compared with that of in low-grade (P>0.05). NAA/Cr and Cho/NAA in tumor parenchyma of high-grade astrocytoma had significant difference compared with that of in low-grade (P<0.01)2. NAA/Cr and Cho/NAA in tumor parenchyma of astrocytoma had significant correlation with pathology grade (P<0.01, respectively).3. When NAA/Cr was 0.51 and Cho/NAA was 2.76, both of their grading results had no significant difference compared with conventional MRI method, but the sensitivity, specificity, PPV,NPV resulted from Cho/NAA were larger and FM, (false positive ratio plus false negative ratio) were lower than conventional MRI method. The grading result of Cho/NAA had significant difference compared with result of NAA/Cr.Conclusions:Both NAA/Cr and Cho/NAA in tumor parenchyma of astrocytoma could be used for grading, while the diagnostic accuracy of NAA/Cr was equal to conventional MRI method and Cho/NAA could improve the diagnostic accuracy for grading. Objective:Our aim was to determine whether the ADC value derived from diffusion weighted imaging (DWI) could be used for suparatentorial glioma grading and assess its diagnostic accuracy.Methods and materials:35 suparatentorial glioma (23 for low-grade,12 for high-grade) confirmed by craniotomy surgery and histopathology underwent conventional MRI plus enhancement and DWI. Measured the minimume ADC value in tumor parenchyma and their contralateral normal brain tissue. Selected the most appropriate cut-off values according to ROC curves. Assessed the correlation of minimum ADC value with pathologic grade of suparatentorial glioma and its diagnostic accuracy for glioma grading. All the data were used SPSS 13.0 for statistics, tested the Homogeneity of variance and Normal distribution before statistics. We used Independent samples t test, Spearman correlation analysis and Fisher’s exact test, and P<0.05 was considered Statistically.Results:1. The ADC value in parenchyma of glioma was significant higher than that in Contralateral normal brain tissue. The ADC value ((1134.05±135.59)×10-6mm2/s) in parenchyma of high-grade glioma was significant lower than that in low-grade glioma ((1472.93±326.52)×10-6mm2/s) (P<0.01).The ADC value in parenchyma of high-grade Oligodendroglial tumors had no difference compared with that in low-grade (P>0.05) but there was significant difference between high-grade astrocytoma and low-grade astrocytoma (P<0.01)2. The ADC value in tumor parenchyma of astrocytoma had significant correlation with pathology grade (P<0.01). The area of ROC curve was 0.922, P<0.01.3. When the ADC value was (1241.05)×10-6mm2/s, its grading results had significant difference compared with conventional MRI method, and the sensitivity, specificity, PPV, NPV resulted from ADC value were larger and FM, (false positive ratio plus false negative ratio) were lower than conventional MRI method.Conclusion:The ADC value in tumor parenchyma of astrocytoma could be used for grading, and its diagnostic accuracy was higher than conventional MRI method which could improve the diagnostic accuracy for grading. Part 3 Superior parameter of multiple MRI method for the assessment of supratentorial glioma gradeObjective:Our aim was to determine which parameter was the best one derived from PWI, SWI, DWI, and 1H MRS that used for the assessment of suparatentorial glioma grading.Methods and materials:32 suparatentorial glioma (21 for low-grade,11 for high-grade) confirmed by craniotomy surgery and histopathology underwent conventional MRI plus enhancement, SWI, PWI, DWI and 1H MRS. According to the degree susceptibility effect (SED) classified the suparatentorial glioma in high-grade and low-grade. Measured the Cho/NAA and mimnmum ADC value in parenchyma of tumor. Calculated the percentage of signal intensity recovery derived from peufusion curve. Selected the most appropriate threshold values of Cho/NAA, ADC value and the percentage of signal intensity recovery according to ROC curves. Assessed the correlation of SED, Cho/NAA, ADC value and the percentage of signal intensity recovery with pathologic grade of suparatentorial glioma respectively and compared their diagnostic accuracy for glioma grading. All the data were used SPSS 13.0 for statistcs, tested the Homogeneity of variance and Normal distribution before statistcs. We used Independent samples t test, Pearson and Spearman correlation analysis and Fisher’s exact test, and P<0.05 was considered Statistically significant. Results:1. SED, Cho/NAA, ADC value and the percentage of signal intensity recovery of tumor parenchyma between low-grade and high-grade gliomas had significant difference (P<0.01).2. SED of tumor parenchyma between low-grade and high-grade Oligodendroglial tumors had significant difference (P<0.05) but other three parameters had no significant difference between low-grade and high-grade Oligodendroglial tumours(P>0.05).3. SED, Cho/NAA, ADC value and the percentage of signal intensity recovery of tumor parenchyma between low-grade and high-grade astrocytoma had significant difference respectively (P<0.01); all of the upper four parameters had significant correlation with pathology grade (P<0.01, respectively).4. Using SED, Cho/NAA, ADC value and the percentage of signal intensity recovery of tumor parenchyma for astrocytoma grading, the sensitivity, specificity, PPV, NPV resulted from ADC value, SED and the percentage of signal intensity recovery were larger and FM, (false positive ratio plus false negative ratio) were lower than conventional MRI method. The grading result between SED and the percentage of signal intensity recovery also had significant difference (P<0.01).Conclusion:Using SED, Cho/NAA, ADC value and the percentage of signal intensity recovery of tumor parenchyma could improve the diagnostic accuracy for grading; the parameters of SED and the percentage of signal intensity recovery had the higher grading efficiency, but the SED had the hightest value for clinical application.更多还原