文献速递神经MR

整合栖息地影像组学、深度学习与传统影像组学的多模态框架预测胶质瘤MGMT基因启动子甲基化:整合模型表现更优

A novel multimodal framework combining habitat radiomics, deep learning, and conventional radiomics for predicting MGMT gene promoter methylation in Glioma: Superior performance of integrated models.
PubMed 文献 · Eur J Radiol · 2025-09-06

本研究基于术前增强T1加权MRI构建整合临床数据、传统影像组学、栖息地影像和深度学习的无创预测框架,用于胶质瘤MGMT基因启动子甲基化术前分层。回顾性纳入加州大学旧金山分校410例和本院102例患者,共构建7个模型。整合模型表现最佳,训练集AUC=0.979(95%CI 0.969-0.990)、F1=0.944,测试集AUC=0.777(0.651-0.904)、F1=0.711;单模态模型中栖息地影像组学最优,训练AUC=0.960(0.954-0.983)、测试AUC=0.724(0.573-0.875)。

要点

  • 整合临床、传统影像组学、栖息地影像和深度学习的模型预测MGMT启动子甲基化表现最佳,测试AUC=0.777。
  • 单模态模型中栖息地影像组学表现最优,训练AUC=0.960、测试AUC=0.724。
  • 栖息地影像组学突显肿瘤异质性在预测MGMT启动子甲基化中的关键作用。

核心结论

  • 所提出的多模态框架显著提升胶质瘤MGMT基因启动子甲基化的术前预测,栖息地影像组学突显肿瘤异质性的关键作用,该方案为胶质瘤个体化管理提供可扩展工具。

研究设计与样本

  • 回顾性研究,纳入加州大学旧金山分校410例和本院102例患者,使用术前增强T1加权MRI(对比剂为钆贝葡胺)构建7个模型;栖息地影像组学特征通过k-means聚类从肿瘤亚区提取,深度学习特征通过3D卷积神经网络获取;模型性能基于AUC、F1分数和决策曲线分析评估。

诊断标准与关键数值

  • 整合临床数据、传统影像组学、栖息地影像特征和深度学习的联合模型训练集AUC=0.979(95%CI 0.969-0.990)、F1=0.944,测试集AUC=0.777(95%CI 0.651-0.904)、F1=0.711。
  • 单模态模型中栖息地影像组学优于其他模型,训练集AUC=0.960(95%CI 0.954-0.983),测试集AUC=0.724(95%CI 0.573-0.875)。

局限性

  • 原文未报告局限性。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

英文原文(摘要)· 点击展开

PURPOSE: The present study aimed to develop a noninvasive predictive framework that integrates clinical data, conventional radiomics, habitat imaging, and deep learning for the preoperative stratification of MGMT gene promoter methylation in glioma.

MATERIALS AND METHODS: This retrospective study included 410 patients from the University of California, San Francisco, USA, and 102 patients from our hospital. Seven models were constructed using preoperative contrast-enhanced T1-weighted MRI with gadobenate dimeglumine as the contrast agent. Habitat radiomics features were extracted from tumor subregions by k-means clustering, while deep learning features were acquired using a 3D convolutional neural network. Model performance was evaluated based on area under the curve (AUC) value, F1-score, and decision curve analysis.

RESULTS: The combined model integrating clinical data, conventional radiomics, habitat imaging features, and deep learning achieved the highest performance (training AUC = 0.979 [95 % CI: 0.969-0.990], F1-score = 0.944; testing AUC = 0.777 [0.651-0.904], F1-score = 0.711). Among the single-modality models, habitat radiomics outperformed the other models (training AUC = 0.960 [0.954-0.983]; testing AUC = 0.724 [0.573-0.875]).

CONCLUSION: The proposed multimodal framework considerably enhances preoperative prediction of MGMT gene promoter methylation, with habitat radiomics highlighting the critical role of tumor heterogeneity. This approach provides a scalable tool for personalized management of glioma.

来源:Eur J Radiol,查看原文。版权归原作者所有。

引用

  1. Zhu FY, Chen WJ, Chen HY, et al. A novel multimodal framework combining habitat radiomics, deep learning, and conventional radiomics for predicting MGMT gene promoter methylation in Glioma: Superior performance of integrated models. Eur J Radiol. 2025. PMID: 40929752. doi:10.1016/j.ejrad.2025.112406

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。