摘要
目的探讨ADC全域灰度直方图对原发性中枢神经系统淋巴瘤(PCNSL)、多形性胶质母细胞瘤(GBM)与单发脑转移瘤(SMT)的鉴别诊断价值。方法收集95例经手术病理证实的脑肿瘤患者,其中PCNSL 38例,GBM 29例,SMT 28例。采用MaZda软件于ADC轴位图像上勾画肿瘤ROI,并进行灰度全域直方图分析,获得9个参数,即均值、变异度、峰度、偏度和第1、10、50、90、99百分位数,比较3种肿瘤间各参数的差异,并采用ROC曲线评价其对3种肿瘤的鉴别诊断效能。结果 PCNSL、GBM、SMT间9个参数总体差异均有统计学意义(P均<0.05),其中第50百分位数鉴别诊断GBM与PCNSL的ROC曲线的AUC最大,为0.90,诊断敏感度为84.21%,特异度为86.21%;GBM与SMT间,均值和第50百分位数的AUC均为0.79,其敏感度均为96.43%,特异度均为55.17%;PCNSL与SMT间,第90和99百分位数的AUC均为0.81,敏感度均为92.86%,特异度均为63.16%。结论直方图分析有助于鉴别PCNSL、GBM和SMT。
Objective To investigate the value of ADC global gray histogram in differential diagnosis of primary central nervous system lymphoma(PCNSL),glioblastoma multiform(GBM)and single brain metastasis(SMT).Methods A total of 95 patients with single brain tumors confirmed by surgery and pathology were collected,including 38 PCNSL,29 GBM and 28 SMT.The MaZda software was used to describe tumor ROI on ADC axial images,and the gray scale global histogram analysis was carried out.Nine parameters were obtained,including mean value,variation,kurtosis,skewness,the first percentile(Perc.01%),the 10 th percentile(Perc.10%),the 50 th percentile(Perc.50%),the 90 th percentile(Perc.90%)and the 99 th percentile(Perc.99%),and the differences of parameters among 3 kinds of tumors were compared.ROC curve was used to evaluate the diagnostic efficacy in differential diagnosis of 3 kinds of tumors.ResultsThere were significant differences of the 9 parameters among PCNSL,GBM and SMT(all P〈0.05).The Perc.50% had the largest AUC value(0.90)of ROC curve in differential diagnosis of GBM and PCNSL,with the sensitivity of 84.21%,and the specificity of 86.21%.AUC of mean and Perc.50% were both 0.79 in diagnosis of GBM and SMT,with the sensitivity of 96.43% and the specificity of 55.17%.In diagnosis of PCNSL and SMT,the AUC of Perc.90% and Perc.99% was both 0.81,with the sensitivity of 92.86%,and the specificity of 63.16%.Conclusion Histogram analysis is helpful to the identification of PCNSL,GBM and SMT.
作者
马桢
程敬亮
任琦
张勇
汪卫建
MA Zhen;CHENG Jingliang;REN Qi;ZHANG Yong;WANG Weijian(Department of MR,the First Affiliated Hospital of Zhengzhou University,Zhengzhou 450052,Chin)
出处
《中国医学影像技术》
CSCD
北大核心
2018年第8期1148-1152,共5页
Chinese Journal of Medical Imaging Technology
关键词
磁共振成像
中枢神经系统
淋巴瘤
胶质母细胞瘤
肿瘤转移
直方图分析
Magnetic resonance imaging
Central nervous system
Lymphoma
Glioblastoma
Neoplasm metastasis
Histogram analysis