分子影像学杂志2025，Vol.48Issue(8)：971-977,7.DOI:10.12122/j.issn.1674-4500.2025.08.08
基于DCE-MRI的瘤内瘤周影像组学模型可有效预测乳腺癌脉管侵犯状态
Intratumoral-peritumoral radiomics model based on dynamic contrast-enhanced MRI can effectively predict the vascular invasion status of breast cancer
摘要
Abstract
Objective To evaluate the value of dynamic contrast-enhanced magnetic resonance imaging(DCE-MRI)based intratumoral-peritumoral radiomics model for prediction of lymphovascular invasion(LVI)in breast cancer.Methods This retrospective study analyzed 153 pathologically confirmed breast cancer patients from Peking University Shenzhen Hospital from January 2019 to September 2024.Patients were stratified into a training set(n=107)and a validation set(n=46)via 7:3 random stratified sampling.Tumor regions of interest were manually delineated on the second post-contrast DCE-MRI phase,with four peritumoral regions(2,4,6,8 mm)generated by concentric expansion.Radiomic features were extracted and filtered for stability.Using the Random Forest classifier,intratumoral and multiple peritumoral models were constructed respectively.The optimal performing peritumoral region was selected and integrated with intratumoral radiomics features to establish an intratumoral-peritumoral combined model.A clinical model was constructed using significant predictors from univariate/multivariate analyses.Finally,a comprehensive model fused radiomic and clinical features.Model performance was assessed using AUC under the ROC curve.Results The 6 mm peritumoral model outperformed other peritumoral configurations.In the training set,AUC for the intratumoral,optimal peritumoral(6 mm),intratumoral-peritumoral,and clinical models were 0.850,0.896,0.916,and 0.787,respectively;corresponding validation AUC were 0.785,0.742,0.819,0.748.The combined model demonstrated superior performance,with AUC of 0.934(training)and 0.827(validation).Conclusion The DCE-MRI based intratumoral-peritumoral radiomics model effectively predicts LVI status in breast cancer,with enhanced accuracy when integrated with clinical features.关键词
乳腺癌/影像组学/瘤内瘤周/脉管侵犯Key words
breast cancer/radiomics/intratumoral-peritumoral/lymphovascular invasion引用本文复制引用
汤朝晖,张哲,杨鹏飞,周勋璋,李圳杰,刘雅洁,成官迅..基于DCE-MRI的瘤内瘤周影像组学模型可有效预测乳腺癌脉管侵犯状态[J].分子影像学杂志,2025,48(8):971-977,7.基金项目
深圳市医疗卫生三名工程项目(SZSM202211036) （SZSM202211036）
