弥漫性胶质瘤MRI多参数空间相关性研究及其在病理分级与异柠檬酸脱氢酶基因分型中的预测价值  

Characteristics of spatial correlation between intratumoral MRI multiparameters in diffuse gliomas and its diagnostic value in grading and genotypes

作　　者：周娴 李建瑞 许强 詹天亮 张志强 Zhou Xian;Li Jianrui;Xu Qiang;Zhan Tianliang;Zhang Zhiqiang(Department of Medical Imaging,Jinling Hospital,Nanjing School Clinical Medicine,Nanjing 210002,China;Health Management Center,Jiangsu Provincial Hospital of Chinese Medicine,Nanjing 210014,China;Department of Radiology,Jinling Hospital,Nanjing School Clinical Medicine,Nanjing 210002,China;Department of Medical Imaging,Nanjing University of Chinese Medicine,Nanjing 210002,China)

机构地区：[1]南京医科大学金陵临床医学院医学影像科,南京210002 [2]江苏省中医院健康管理中心,南京200014 [3]南京医科大学金陵临床医学院放射诊断科,南京210002 [4]南京中医药大学金陵临床医学院医学影像科,南京210002

出　　处：《中华解剖与临床杂志》2025年第3期145-153,共9页Chinese Journal of Anatomy and Clinics

基　　金：国家重点研发计划(2018YFA0701703);国家自然科学基金(81530054)。

摘　　要：目的分析弥漫性胶质瘤MRI多序列参数的空间相关性特征,并探讨空间相关性对胶质瘤病理分级及异柠檬酸脱氢酶(IDH)基因分型的预测效能。方法横断面研究。回顾性分析2017年1月—2020年1月南京医科大学金陵临床医学院经病理证实的198例弥漫性胶质瘤患者的多序列MRI影像资料。MRI序列有T1加权像(T1WI)、T2WI、T1加权对比增强(T1CE)、动脉自旋标记(ASL)和弥散加权成像(DWI)。各序列肿瘤实质区信号采用美国MathWorks公司MATLAB R2018B软件进行两两Pearson相关分析,并对相关系数r值进行Fisher Z转换。定量分析肿瘤实性部分多序列指标[T1CE、T1WI及T2WI信号强度,表观弥散系数(ADC)值,以及脑血流量(CBF)值]间的空间相关性特征。分析多指标间空间相关性特征与病理分级及IDH基因分型的关系。利用支持向量机(SVM)分类器建立MRI多指标空间相关性特征模型,用于胶质瘤病理分级及IDH基因分型的预测。结果198例患者中,男109例、女89例,年龄23~80(52.3±13.3)岁;低级别胶质瘤(LGG)组71例(Ⅱ级)、高级别胶质瘤HGG组127例(Ⅲ级42例、Ⅳ级85例)。IDH基因型检测的145例中,IDH野生型组89例、IDH突变型组56例。全组198例患者中MRI多序列指标相关性分析显示,T1CE与CBF、T1CE与T1WI、ADC与T2WI、CBF与T2WI指标呈两两正相关,且ADC与T2WI正相关性最强;T1CE与ADC、T1CE与T2WI、ADC与CBF、ADC与T1WI、CBF与T1WI、T1WI与T2WI指标呈两两负相关,且ADC与T1WI负相关性最强。检测IDH基因分型的145例患者中,除CBF与T2WI指标呈两两负相关外,其他各指标间的相关性方向与之前一致,且T1WI与T2WI负相关性最强。在HGG与LGG组中MRI的T1CE与ADC、T1CE与CBF、T1CE与T1WI、T1CE与T2WI、ADC与CBF、ADC与T1WI、CBF与T1WI、CBF与T2WI、T1WI与T2WI的空间相关性差异均有统计学意义(P值均<0.05),且基于此9项指标间空间相关性构建的SVM预测模型的验证集结果显示其有较高的病理�Objective This study aimed to analyze the spatial correlation characteristics of multisequence MRI indicators in diffuse gliomas and to investigate their predictive efficacy for tumor pathological grading and isocitrate dehydrogenase(IDH)genotypes.Methods A cross-sectional study was conducted to retrospectively analyze the multisequence MRI imaging data of 198 patients with pathologically confirmed diffuse gliomas at the Jinling Hospital,Nanjing School Clinical Medicine from January 2017 to January 2020.The MRI sequences included T1-weighted imaging(T1WI),T2-weighted imaging(T2WI),T1-weighted contrast-enhanced(T1CE),arterial spin labeling(ASL),and diffusion-weighted imaging(DWI).The signals from the tumor solid areas in each sequence were quantitatively analyzed using MATLAB R2018B software from MathWorks,USA,and pairwise Pearson correlation analysis was performed.The correlation coefficients(r values)were then subjected to Fisher Z transformation.The spatial correlation characteristics between multisequence indicators of the tumor solid part(T1CE,T1WI,and T2WI relative signal intensity,apparent diffusion coefficient[ADC]value,and cerebral blood flow[CBF]value)were examined.Additionally,the relationship between these spatial correlation characteristics and pathological grading and IDH mutation type was analyzed.A support vector machine(SVM)classifier was used to establish a model based on MRI multi-indicator spatial correlation features for predicting glioma pathological grading and IDH mutation type.Results Among the 198 patients,109 were male and 89 were female,with ages ranging from 23 years to 80 years(mean 52.3±13.3 years).The cohort included 71 low-grade gliomas(LGG,all gradeⅡ)and 127 high-grade gliomas(HGG,42 gradeⅢ,and 85 gradeⅣ).Of the 145 patients tested for IDH gene type,89 were classified as IDH wild types and 56 as IDH mutants.In this study,positive correlations were observed between T1CE and CBF,T1CE and T1WI,ADC and T2WI,and CBF and T2WI,with the strongest positive correlation found between

关 键 词：神经胶质瘤 多参数磁共振成像 空间相关性 组织学分级 异柠檬酸脱氢酶 

分 类 号：R739.4[医药卫生—肿瘤]