DCE-MRI影像组学联合临床特征鉴别luminal型与非luminal型乳腺癌的价值
Value of DCE-MRI Imaging and Combined Clinical Features in Differentiating Luminal and Non-luminal Breast Cancer
贾燕茹 1柴军 2赵建华 2刘宇 1宋丹 1薛瑞红 1张煜杰 3王晓越4
作者信息
- 1. 内蒙古科技大学包头医学院(内蒙古包头 014010)
- 2. 内蒙古自治区人民医院影像医学科(内蒙古呼和浩特 010017)
- 3. 内蒙古大学电子信息工程学院(内蒙古呼和浩特 010021)
- 4. 燕山大学电气工程学院(河北秦皇岛 066004)
- 折叠
摘要
目的 探讨基于DCE-MRI影像组学联合临床特征在术前鉴别诊断luminal型与非luminal型乳腺癌的价值.方法 回顾性纳入2014年1月～2023年8月内蒙古自治区人民医院行DCE-MRI检查并获得病理结果的212例乳腺癌患者的图像,根据病理分子分型,将患者分为luminal型乳腺癌114例,非luminal型98例.按7:3的比例148例为训练组,64例为验证组.采用3D-slicer软件手动勾画病灶体积兴趣区(VOI),经LASSO回归及T检验批量提取影像组学特征并进行筛选,筛选出有价值的影像组学特征及临床特征,分别构建2个预测模型:单一影像组学模型及影像组学联合临床特征模型.采用ROC曲线下面积、准确率、敏感度、特异度以及校准曲线评价训练集中影像组学联合临床特征鉴别luminal型与非luminal型乳腺癌的诊断效能.结果 单一影像组学模型的AUC值0.850、准确率0.790、敏感度0.820、特异度0.765;影像组学联合临床特征模型AUC值0.856、准确率0.783、敏感度0.791、特异度0.776.临床特征中,如绝经状态(P=0.009)、淋巴结转移(P=0.012)有统计学意义,年龄(P=0.165)、病理类型(P=0.687)无统计学意义.结论 基于DCE-MRI的影像组学联合临床特征在术前鉴别luminal型与非luminal型乳腺癌方面具有较大价值.
Abstract
Objective To explore the value of DCE-MRI radiomics combined with clinical features in the preoperative differential diagnosis of luminal and non-Luminal breast cancer.Methods The images of 212 patients with breast cancer who underwent DCE-MRI examination and obtained pathological results from Inner Mongolia Autonomous Region People's Hospital from January 2014 to August 2023 were retrospectively analyzed.According to the pathological molecular classification,114 cases of luminal breast cancer and 98 cases of non-Luminal breast cancer were classified.According to the ratio of 7:3,148 cases were training group and 64 cases were verification group.3D-slicer software was used to manually delineate the focal volume area of interest(VOI),and the imaging omics features were extracted and screened in batches by LASSO regression and T test.Valuable imaging omics features and clinical features were screened out,and two prediction models were constructed:pure radiomics model and combined radiomics clinical feature model.The area under ROC curve,accuracy,sensitivity,specificity and calibration curve were used to evaluate the diagnostic efficiency of luminal and non-Luminal breast cancer by the imaging features and combined clinical features of the training set.Results AUC value of pure radiomics model was 0.850,accuracy 0.790,sensitivity 0.820 and specificity 0.765.The AUC value of radiomic features combined with clinical features was 0.856,the accuracy was 0.783,the sensitivity was 0.791,and the specificity was 0.776.Clinical features such as menopausal status(P=0.009)and lymph node metastasis(P=0.012)had statistical significance,while age(P=0.165)and pathological type(P=0.687)had no statistical significance.Conclusion The radiomic features and combined clinical features based on DCE-MRI are of great value in differentiating luminal and non-luminal breast cancer before surgery.
关键词
影像组学/乳腺癌/分子分型/磁共振成像Key words
Radiomics/Breast Cancer/Molecular Typing/MRI引用本文复制引用
基金项目
内蒙古自治区-上海交通大学科技合作专项"科技兴蒙"上海交通大学行动计划子项目(2022XYJG0001)
内蒙古自治区人民医院院内项目(2020YN08)
包头医学院研究生教育教学改革项目(B-YJSJG202303)
内蒙古医科大学高等教育教学改革研究项目(2023)(NYJXGG2023139)
内蒙古医科大学联合项目(YKD2023LH088)
出版年
2024
NETL
NSTL
万方数据