基于常规MRI影像组学无创性诊断胶质细胞瘤IDH1基因型的研究  

A Study Based on Conventional MRI Radiomics for Non-invasive Diagnosis of Glioblastoma IDH1 Genotype

作　　者：徐慰幸子 周维燕 文剑波 黎元 XU Wei-xing-zi;Zhou Wei-yan;Wen Jian-bo;Li Yuan(Department of Radiology,Zhongshan Hospital,Fudan University,Shanghai 200030,China;PET Center,Huashan Hospital,Fudan University,Shanghai 200040,China;Department of Radiology,Huashan Hospital,Fudan University,Shanghai 200040,China)

机构地区：[1]复旦大学附属中山医院放射科,上海200030 [2]复旦大学附属华山医院PET中心,上海200040 [3]复旦大学附属华山医院放射科,上海200040

出　　处：《中国CT和MRI杂志》2023年第10期1-4,共4页Chinese Journal of CT and MRI

摘　　要：目的探讨胶质细胞瘤IDH1基因突变型与野生型的常规MRI特征,评价常规影像对胶质瘤IDH1基因状态的无创性预测的价值。方法回顾性分析2014年1月至2018年12月期间华山医院胶质瘤共305例。收集强化均匀性等24个特征。根据2018版WHQ胶质瘤病理分级,将入组患者分为低级别(WHOⅠ和Ⅱ级)胶质瘤I DH1基因野生型与突变型组;高级别(WHOⅢ和Ⅳ级)胶质瘤IDH1基因野生型与突变型四组。采用Feature Selection Library(FSLib)中的mrmr等十种特征选择方法,筛出与IDH1基因相关的影像学及临床特征,计算特征的P值。结果在低级别胶质瘤组中,选择的6个有效特征集为(年龄、脑干、块状强化、左侧额叶、岛叶、T_(2)WI信号),诊断正确率为79%。在高级别胶质瘤组中,筛出的6个有效特征集为(强化程度、水肿均匀性、坏死、强化均匀性、环形强化、钙化),诊断正确率为8.2%。结论基于常规影像学特征选择数据分析有助于术前无创性诊断胶质瘤IDH1基因状态。Objective To investigate the conventional MRI characteristics of glioblastoma IDH1 gene mutation type and wild type,and to evaluate the value of conventional imaging in the noninvasive prediction of glioma isocitrate dehydrogenase 1(IDH1)gene status.Methods A total of 305 glioma patients in Huashan Hospital were retrospectively analyzed from January 1,2014 to December 31,2018.Collection and collation:24 characteristics such as degree of edema and uniformity of reinforcement.According to the pathological classification of glioma of the World Health Organization in 2016,the enrolled patients were divided into low-grade(WHO gradeⅠandⅡ)glioma IDH1 gene wild type and mutant group;High-grade(WHO gradeⅢandⅣ)glioma IDH1 gene wild-type and mutant four groups.The common methods of ten feature selections,such as mrmr and relieff in the Feature Selection Library(FSLib),were used to screen out the conventional imaging and general clinical features related to the IDH1 gene,and the p-value of the selected features was calculated.Results In the low-grade glioma group,the six effective feature sets selected(age,brainstem,massive enhancement,left frontal lobe,insular lobe,T_(2)WI signal)were selected with an average diagnostic accuracy rate of 79%.In the high-grade glioma group,the six effective feature sets screened out(degree of intensification,uniformity of edema,necrosis,uniformity of reinforcement,ring enhancement,calcification),and the average diagnostic accuracy rate was 82%.Conclusion Data analysis based on conventional imaging feature selection is helpful for the non-invasive diagnosis of IDH1 gene status of glioma before surgery.

关 键 词：胶质细胞瘤 影像特征集 异柠檬酸盐脱氢酶1(IDH1) 特征集选择 

分 类 号：R73[医药卫生—肿瘤]