应用受试者工作特征曲线分析多模态MRI对原发性中枢神经系统淋巴瘤与胶质母细胞瘤鉴别诊断价值  

Value of Multi-Modal MRI in Differential Diagnosis of Primary Central Nervous System Lymphoma and Glioblastoma by Receiver Operating Characteristic Curve

作　　者：宋伟[1] 邓雪飞[2] 王肖 徐茂林 谭德力 骆祥伟[1] 朱友志[1] 张禹[1] SONG Wei;DENG Xuefei;WANG Xiao(Department of Medical Imaging,The 901st Hospital of the Joint Logistics Support Force of PLA,Hefei 230031,P.R.China)

机构地区：[1]中国人民解放军联勤保障部队第901医院医学影像科,230031 [2]安徽医科大学解剖教研室

出　　处：《临床放射学杂志》2019年第11期2026-2031,共6页Journal of Clinical Radiology

基　　金：安徽省卫生和计划生育委员会科研计划项目(编号:2016QK080);安徽省重点研究与开发计划项目(编号:1804h08020284)。

摘　　要：目的探讨多模态MRI对原发性中枢神经系统淋巴瘤(PCNSL)和胶质母细胞瘤(GBM)的鉴别价值。方法回顾性分析经病理证实的20例PCNSL和25例GBM患者资料。患者术前均行扩散加权成像(DWI)、磁敏感加权成像(SWI)、动态磁敏感对比增强灌注成像(DSC-PWI),测肿瘤平均表观扩散系数值(ADCmean)、磁敏感信号强度(ITSS)分级、脑血容量平均值比值(rCBVmean)。用独立样本t检验比较两者多模态MRI参量差异;用受试者工作特征曲线(ROC)评价各参量诊断效能,获取最佳阈值、敏感度、特异度、准确率等指标;用Z检验比较各参量间ROC曲线下面积(AUC)差异。结果PCNSL组ADCmean平均值(0.64±0.09)×10-3mm2/s,ITSS平均值(0.75±0.64)级,rCBVmean平均值1.67±0.77;GBM组ADCmean平均值(0.86±0.16)×10-3mm2/s,ITSS平均值(2.56±0.58)级,rCBVmean平均值3.53±0.82;各参量均有统计学差异(P均<0.001)。ROC提示ADCmean最佳阈值0.84×10-3mm2/s,AUC 0.902,敏感度100%,特异度72%,准确率72%;ITSS最佳阈值1级,AUC 0.967,敏感度90%,特异度96%,准确率86%;rCBVmean最佳阈值2.69,AUC 0.942,敏感度95%,特异度92%,准确率87%;ADCmean与ITSS分级及rCBVmean联合诊断效能最优(AUC 0.996,敏感度100%,特异度96%,准确率96%),其AUC与ADCmean间有差异(P=0.038),与双参量及其他单参量间无差异;双参量AUC与单参量相比,仅ADCmean+ITSS、ITSS+rCBVmean与ADCmean(P=0.037、0.038)间有差异;双参量间AUC无差异。结论基于平扫的DWI和SWI联合对PCNSL与GBM鉴别能力与多模态MRI联合相似,且优于单模态MRI,有较高临床实用价值。Objective To investigate the diagnostic value of multi-modal MRI in differentiating primary central nervous system lymphoma(PCNSL) and glioblastoma(GBM) through receiver operating characteristic(ROC) curve. Methods 20 patients with PCNSL and 25 patients with GBM confirmed by pathology were analyzed retrospectively. All patients underwent diffusion weighted imaging(DWI),susceptibility weighted imaging(SWI) and dynamic susceptibility contrast perfusion weighted imaging(DSC-PWI) preoperatively. The mean apparent diffusion coefficient(ADCmean) of tumorous parenchyma was measured;intratumoral susceptibility signal intensity(ITSS)of that was graded;ratio of mean value of cerebral blood volume(rCBVmean) was calculated. Differences of multi-modal MRI parameters were statistically inspected through independent sample t-test. ROC curve was used to evaluate diagnostic efficiency,in order to obtain threshold,sensitivity,specificity,accuracy and Z test was used to compare the difference of area under the ROC curve(AUC) between various parameters. Results There were statistically significant difference in the ADCmean [(0.64±0.09)×10-3mm2/s of PCNSL,(0.86±0.16)×10-3mm2/s of GBM],grade of ITSS(0.75±0.64 grads of PCNSL,2.56±0.58 grads of GBM),rCBVmean(1.67±0.77 of PCNSL,3.53±0.82 of GBM) between two tumors(P<0.001,P<0.001,P<0.001). ROC curve revealed that setting ADCmean=0.84×10-3mm2/s as threshold value,optimal diagnostic ability could be obtained(AUC:0.902,sensitivity:100%,specificity:72%,accuracy:72%);setting ITSS=1 as threshold value,optimal diagnostic ability of PCNSL could be obtained(AUC:0.967,sensitivity:90%,specificity:96%,accuracy:86%);setting rCBVmean=2.69 as threshold value,optimal ability could be obtained(AUC:0.942,sensitivity:95%,specificity:92%,accuracy:87%). The combination of ADCmean and ITSS and rCBVmean had best diagnostic efficiency(AUC:0.996,sensitivity:100%,specificity:96%,accuracy:96%). Comparison of the differences in AUC between combination of three parameters and any combination of double param

关 键 词：原发性中枢神经系统淋巴瘤 胶质母细胞瘤 扩散加权成像 灌注加权成像 磁敏感加权成像 受试者工作特征曲线 

分 类 号：R445.2[医药卫生—影像医学与核医学] R739.4[医药卫生—诊断学]