基于自动乳腺容积扫查影像组学列线图预测浸润性乳腺癌腋窝淋巴结风险的研究  

Research of Predicting Axillary Lymph Node Risk in Invasive Breast Cancer Based on Automated Breast Volume Scanner Nomogram

作　　者：王惠[1] 马苏美[1] 陈飞[1] 雷军强[2] 南彩玲[1] 张炜阳[1] 郭顺林[2] Wang Hui;Ma Sumei;Chen Fei;Lei Junqiang;Nan Cailing;Zhang Weiyang;Guo Shunlin(Department of Ultrasound,The First Hospital of Lanzhou University,Lanzhou 730000,China;Department of Radiology,The First Hospital of Lanzhou University,Lanzhou 730000,China)

机构地区：[1]兰州大学第一医院超声科,兰州市730000 [2]兰州大学第一医院放射科,兰州市730000

出　　处：《中国超声医学杂志》2022年第7期762-766,共5页Chinese Journal of Ultrasound in Medicine

基　　金：兰州大学第一医院院内基金(No.ldyyyn2019-65)。

摘　　要：目的 本研究旨在联合乳腺癌自动乳腺容积扫描(ABVS)的影像组学特征和临床及病理因素构建列线图,以预测浸润性乳腺癌患者腋窝淋巴结转移风险。方法 回顾性研究102例浸润性乳腺癌患者的自动乳腺容积扫查图像,分为训练集和验证集。应用类内相关系数评估影像组学特征一致性,采用最小绝对收缩和选择算子逻辑回归建立预测腋窝淋巴结转移的影像组学标签。利用单因素逻辑回归,从自动乳腺容积扫查影像组学标签和临床、病理因子,筛选出腋窝淋巴结转移的显著预测因子。采用多因素逻辑回归,进一步筛选出腋窝淋巴结转移的独立预测因子,并构建列线图。采用受试者工作特征曲线评估列线图预测性能。结果 列线图包括影像组学标签和增殖细胞核抗原Ki67。在训练集和验证集中,列线图对腋窝淋巴结转移具有中等预测效能,受试者工作特征曲线下面积分别为0.81[95%可信区间(CI):0.70~0.91]、0.75(95%CI:0.57~0.92)。结论 基于ABVS和Ki67的影像组学列线图是准确预测腋窝淋巴结转移和优化浸润性乳腺癌临床决策的无创工具。Objective We aimed to develop a nomogram including radiomics features of automated breast volume scanner(ABVS) and clinicopathological factors to evaluate the axillary lymph node metastases risk in patients with invasive breast cancer. Methods We retrospectively analyzed 102 ABVS images from patients with invasive breast cancer and divided into training and validation sets. The intraclass correlation coefficient was utilized to assess the consistency of the radiomics features. The radiomics signature was established with the least absolute shrinkage and selection operator. Univariate logistic regression was utilized to screen for significant predictors of axillary lymph node metastasis from ABVS based radiomics signature and clinicopathological factors. Multivariate logistic regression was used to further screen independent predictors and construct a nomogram. The receiver operating characteristic curve was utilized to assess the nomogram’s predictive performance. Results The nomogram model was composed of ABVS radiomics features and Proliferating cell nuclear antigen Ki67, achieved medium predictive efficiency, and the area under the receiver operating characteristic curve [95% confidence interval(CI)] were 0.81(0.70~0.91) and 0.75(0.57~0.92), respectively. Conclusions ABVS and Ki67-based radiomics nomogram were a non-invasive tool to accurately predict axillary lymph node metastasis and optimize clinical decision making of invasive breast cancer.

关 键 词：乳腺癌 自动乳腺容积扫描 列线图 腋窝淋巴结转移 

分 类 号：R445.1[医药卫生—影像医学与核医学] R737.9[医药卫生—诊断学]