多模态MRI列线图模型预测非典型脑膜瘤瘤周脑组织侵犯  

Prediction of Peritumoral Brain Tissue Invasion in Atypical Meningioma by Multimodality MRI-Based Nomogram Model

作　　者：赵益晶 胡建平[1] 李猛城 曹代荣[1] 邢振[1] ZHAO Yijing;HU Jianping;LI Mengcheng;CAO Dairong;XING Zhen(Department of Radiology,the First Affiliated Hospital Fujian Medical University,Fuzhou 350004,China)

机构地区：[1]福建医科大学附属第一医院影像科,福建福州350004

出　　处：《中国医学影像学杂志》2025年第1期12-17,47,共7页Chinese Journal of Medical Imaging

基　　金：国家自然科学基金面上项目(82071869,82371905)。

摘　　要：目的评估多模态MRI列线图模型预测非典型脑膜瘤瘤周脑组织侵犯的价值。资料与方法回顾性纳入2018年1月—2023年1月福建医科大学附属第一医院187例经术后病理证实的非典型脑膜瘤患者,其中瘤周脑组织侵犯130例,无瘤周脑组织侵犯57例。收集患者临床资料及多模态MRI参数,包括年龄、性别、肿瘤部位、肿瘤最大径、瘤周水肿范围、瘤-脑界面模糊、分叶征、脑膜尾征、囊变/坏死、相对最小表观扩散系数、肿瘤内磁敏感信号分级,比较两组间参数差异,并应用单因素和多因素Logistic回归分析筛选非典型脑膜瘤瘤周脑组织侵犯的独立预测因素,应用Logistic回归构建多模态MRI预测模型,并绘制列线图。应用受试者工作特征曲线评估多模态MRI列线图模型及各独立预测因素的效能。结果肿瘤最大径(OR=0.705,95%CI 0.539~0.920,P=0.010)、瘤周水肿范围(OR=1.333,95%CI 1.095~1.624,P=0.004)、瘤-脑界面模糊(OR=5.121,95%CI 2.045~12.826,P<0.001)及相对最小表观扩散系数(OR=0.126,95%CI 0.033~0.483,P=0.002)为非典型脑膜瘤瘤周脑组织侵犯的独立预测因素;基于此构建的多模态MRI列线图获得曲线下面积为0.80(95%CI 0.73~0.88),相应的敏感度、特异度分别为87.69%、66.67%;其曲线下面积大于肿瘤最大径、瘤周水肿范围、瘤-脑界面模糊及相对最小表观扩散系数的曲线下面积(Z=3.665、3.904、4.359、3.701,P均<0.05)。结论多模态MRI列线图模型有助于预测非典型脑膜瘤瘤周脑组织侵犯。Purpose To evaluate the value of multimodality MRI-based nomogram model for predicting peritumoral brain tissue invasion in atypical meningioma.Materials and Methods A total of 187 patients with pathologically diagnosed atypical meningioma in the First Affiliated Hospital Fujian Medical University from January 2018 to January 2023 were retrospectively enrolled,including 130 cases of peritumoral brain tissue invasion and 57 cases of none peritumoral brain tissue invasion.Clinical data and multimodality MRI features,including age,gender,tumor location,maximum diameter,peritumoral oedema,tumor-brain interface,lobulated sign,dural tail sign,cyst degeneration/necrosis,relative minimum apparent diffusion coefficient(rADCmin)and intratumoral susceptibility signal were analyzed.Univariate and multivariate Logistic regression analysis were performed to screen the independent predictors of peritumoral brain tissue invasion in atypical meningioma,then a multimodality MRI prediction model was constructed,and was visualized as a nomogram.The prediction performance of multimodality MRI-based nomogram model and each independent predictor were assessed using receiver operating characteristic curve.Results The maximum diameter(OR=0.705,95%CI 0.539-0.920,P=0.010),peritumoral oedema(OR=1.333,95%CI 1.095-1.624,P=0.004),tumor-brain interface(OR=5.121,95%CI 2.045-12.806,P<0.001)and rADCmin(OR=0.126,95%CI 0.033-0.483,P=0.002)were independent predictors of peritumoral brain tissue invasion in atypical meningioma.The area under the curve,sensitivity and specificity of the multimodality MRI-based nomogram model for predicting peritumoral brain tissue invasion in atypical meningioma was 0.80(95%CI 0.73-0.88),87.69%and 66.67%,respectively.The multimodality MRI-based nomogram model showed significantly higher area under the curve than that of the maximum diameter,peritumoral oedema,tumor-brain interface and rADCmin of atypical meningioma(all Z=3.665,3.904,4.359,3.701,P<0.05).Conclusion The multimodality MRI-based nomogram model may be helpfu

关 键 词：非典型脑膜瘤 脑组织侵犯 磁共振成像 列线图表 影像组学 预测 

分 类 号：R739.45[医药卫生—肿瘤] R445.2[医药卫生—临床医学]