MRI影像组学预测乳腺癌腋窝淋巴结转移
MRI Radiomics Predicts ALN in Breast Cancer
赖木雅 1刘子霖 1谭学渊 1农盛1
作者信息
- 1. 广东医科大学附属第二医院放射影像科(524000)
- 折叠
摘要
目的:旨在评估动态增强磁共振成像(DCE-MRI)影像组学模型在预测乳腺癌患者腋窝淋巴结转移方面的有效性.方法:本研究基于2021年1月至2023年11月期间接受腋窝淋巴结清扫或活检的63名乳腺癌患者,构建了一个影像组学模型.我们从每位患者DCE-MRI中增强第3期的图像中提取肿瘤相关的影像组学特征.然后使用最小绝对收缩和选择算子(LASSO)方法来筛选出有预测价值的特征,并利用随机森林算法构建预测腋窝淋巴结转移的影像组学模型,然后和临床影像特征联合构建联合模型.同时,通过单因素和多因素逻辑回归分析确定与腋窝淋巴结转移相关的临床危险因素.在此基础上,运用多因素逻辑回归分析将其影像组学评分与临床危险因素结合,形成一个综合的临床-影像组学评分模型.结果:最终筛选出8个影像组学特征和一个临床特征用于预测乳腺癌腋窝淋巴结转移.综合模型的预测性能为0.8,效果优于单纯基于影像组学特征的模型.结论:基于DCE-MRI的影像组学模型可以在术前无创评估腋窝淋巴结转移情况,可以为临床提供一种强大的预测工具.
Abstract
Objective:The aim was to assess the validity of a dynamic-enhanced magnetic resonance imaging(DCE-MRI)radiomics model in predicting axillary lymph node metastasis(ALN)in breast cancer patients.Methods:In this study,an radiomics model was constructed based on 63 breast cancer patients who underwent axillary lymph node dissection or biopsy between January 2021 and November 2023.We extracted tumor-related radiomics features from the enhanced phase 3 images in DCE-MRI of each patient.Then we used the least absolute shrinkage and selection operator(LASSO)method to filter out the features with predictive value,and constructed an imaging histology model for predicting ALN using the random forest algorithm,and then constructed a joint model in conjunction with the clinical imaging features.At the same time,clinical risk factors associated with ALN were determined by single-factor and multifactor logistic regression analysis.On this basis,a multifactorial logistic regression analysis was applied to combine their imaging histology scores with clinical risk factors to form a comprehensive clinical-imaging histology score model.Results:Eight imaging histologic features and one clinical feature were finally screened for predicting ALN in breast cancer.The predictive performance of the integrated model was 0.8,which was better than the model based on imaging histologic features alone.Conclusion:The DCE-MRI-based radiomics model can be used to assess ALN noninvasively in the preoperative period,and it can provide a powerful predictive tool in the clinic.
关键词
MRI影像组学/乳腺癌/腋窝淋巴结转移Key words
MRI radiomics/breast cancer/axillary lymph node metastasis引用本文复制引用
出版年
2024
NETL
NSTL
万方数据