基于弥散峰度成像影像组学列线图预测乳腺癌脉管浸润  

Prediction of the Lymphovascular Invasion in Breast Cancer Based on Diffusion Kurtosis Imaging Derived Radiomics Nomogram

作　　者：张思琦 潘海宾 郁伟斌 蒋璟璇 李跃华[2] ZHANG Siqi;PAN Haibin;YU Weibin;JIANG Jingxuan;LI Yuehua(Department of Imaging,Affiliated Hospital of Nantong University,Nantong 226001,China;Institute of Diagnostic and Interventional Radiology,Shanghai Sixth People's Hospital;Department of Radiology,Shanghai Sixth People's Hospital South Campus Fengxian District Central Hospital)

机构地区：[1]南通大学附属医院影像科,南通226001 [2]上海市第六人民医院放射介入科 [3]上海市第六人民医院南院奉贤区中心医院放射科

出　　处：《中国医学计算机成像杂志》2024年第2期156-160,共5页Chinese Computed Medical Imaging

基　　金：奉贤区科委社会类科技发展基金项目(奉科20201417)。

摘　　要：目的:基于磁共振弥散峰度成像(DKI)序列参数图像构建影像组学列线图模型,对乳腺癌脉管浸润进行术前预测。方法:纳入2018年12月至2022年12月术前行DKI检查的乳腺癌患者169例,根据术后病理将其分为浸润组(73例)和非浸润组(96例),按照7︰3比例分为训练集(118例,浸润49例)和验证集(51例,浸润24例)。基于DKI参数图像提取肿块的影像组学特征,构建影像组学模型。采用逻辑回归联合影像组学和临床特征构建列线图模型,并绘制受试者工作特征(ROC)曲线以评估各模型的诊断效能。结果:在测试队列中,列线图模型的曲线下面积明显高于临床模型(0.864 vs 0.725,P<0.05)。结论:使用基于DKI提取影像组学联合临床特征构建的列线图模型,可以有效预测乳腺癌患者术前的脉管浸润情况。Purpose:To predict preoperative lymphovascular invasion in breast cancer patients by constructing a radiomics nomogram model based on diffusion kurtosis imaging(DKI)sequence parameter images.Methods:A total of 169 breast cancer patients who underwent preoperative DKI examinations from December 2018 to December 2022 were included and classified into invasive(73 cases)and non-invasive(96 cases)groups according to postoperative pathology.The sample was divided into a training set(118 cases,49 invasive)and a validation set(51cases,24 invasive)at a ratio of 7:3.Radiomics features of the lesions were extracted from DKI parameter images to build the radiomics model.A nomogram model was constructed by combining radiomics and clinical features using logistic regression,and receiver operating characteristic(ROC)curves were plotted to evaluate the diagnostic performance of the models.Results:In the test cohort,the area under the curve of the nomogram model was significantly higher than that of the clinical model(0.864 vs 0.725,P<0.05).Conclusion:The nomogram model constructed by combining radiomics features extracted from DKI with clinical features can effectively predict preoperative lymphovascular invasion in breast cancer patients.

关 键 词：弥散峰度成像 影像组学 乳腺癌 脉管浸润 列线图 

分 类 号：R730.49[医药卫生—肿瘤]