弥漫性胶质瘤患者生存预后评估——基于MRI影像组学特征  

Survival Prognosis Assessment Study in Patients with Diffuse Glioma:Based on MRI Radiomics Features

作　　者：叶大涵 谢剑锋[2,3,4] 陈晓平 YE Dahan;XIE Jianfeng;CHEN Xiaoping(School of Mathematics and Statistics,Fujian Normal University,Fuzhou 350117,China;Fujian Provincial Center for Disease Control and Prevention,Fuzhou 350012,China;Fujian Provincial Key Laboratory of Zoonosis Research,Fuzhou 350012,China;Practice Base of Public Health School,Fujian Medical University,Fuzhou 350012,China;School of Computer Science and Mathematics,Fujian University of Technology,Fuzhou 350118,China;Fujian Provincial Key Laboratory of Statistics and Artificial Intelligence(Fujian Normal University),Fuzhou 350117,China)

机构地区：[1]福建师范大学数学与统计学院,福建福州350117 [2]福建省疾病预防控制中心,福建福州350012 [3]福建省人兽共患病重点实验室,福建福州350012 [4]福建医科大学公共卫生学院实践基地,福建福州350012 [5]福建理工大学计算机科学与数学学院,福建福州350118 [6]统计学与人工智能福建省高校重点实验室(福建师范大学),福建福州350117

出　　处：《福建师范大学学报(自然科学版)》2025年第2期9-15,共7页Journal of Fujian Normal University:Natural Science Edition

基　　金：国家自然科学基金项目(12401380);福建省卫生健康重大科研专项(2021ZD01001);福建省直单位教育和科研专项资金(闽财指[2022]639号);福建省卫生健康中青年领军人才培养项目;福建省教育科学规划2024年教育考试招生重点专项课题(FJJKS24-02,福建省教育考试院资助);福建理工大学科研启动基金(GY-S24002)。

摘　　要：基于T1加权后对比增强MRI的影像组学特征,开发并验证一个用于预测弥漫性胶质瘤患者术后生存预后的模型。从387例弥漫性胶质瘤患者的术前MRI数据中提取出坏死肿瘤核心区域、瘤周水肿区域以及增强肿瘤区域的影像组学特征,并使用LASSO-Cox回归模型进行特征选择与模型构建。数据被随机分为训练组和验证组,模型在训练组患者(80%;309例)中开发,并在验证组患者(20%;78例)中验证预测性能。模型在训练组和验证组中的C-index分别为0.72和0.69;在训练组中1年期、2年期、3年期的AUC值分别为0.79、0.81、0.86,而在验证组中,对应的AUC值为0.72、0.76、0.81。研究表明,基于T1加权后对比增强MRI的影像组学特征,可有效预测弥漫性胶质瘤患者的术后预后,并为临床决策提供依据。A model based on radiomics features from contrast-enhanced T1-weighted MRI was developed and validated to predict postoperative survival prognosis in patients with diffuse glioma.Radiomics features were extracted from the necrotic tumor core necrosis,peritumoral edema and enhancing tumor regions using preoperative MRI data from 387 patients with diffuse glioma,and a LASSO-Cox regression model was employed for feature selection and model construction.The data were randomly divided into a training group(80%;309 patients)and a validation group(20%;78 patients).The model was developed in the training group and its predictive performance was validated in the validation group.The model's C-index was 0.72 in the training group and 0.69 in the validation group.The 1-year,2-year,and 3-year AUCs were 0.79,0.81,and 0.86 in the training group,and 0.72,0.76,and 0.81 in the validation group,respectively.The study demonstrated that radiomics features based on contrast-enhanced T1-weighted MRI can effectively predict the postoperative prognosis of patients with diffuse gliomas,providing a basis for clinical decision-making.

关 键 词：弥漫性胶质瘤 磁共振成像 影像组学 生存分析 预后 LASSO-Cox回归模型 

分 类 号：O213[理学—概率论与数理统计]