Prediction of Axillary Lymph Node Metastasis Based on Intratumoral and Peritumoral Ultrasound Radiomics Features of the Primary Lesion of Breast Cancer
10.3969/j.issn.1005-5185.2025.10.006
- VernacularTitle:基于乳腺癌原发灶瘤内及瘤周超声影像组学特征预测腋窝淋巴结转移
- Author:
Yao DU
1
;
Meng WU
;
Yuhua WANG
;
Xiaodan FENG
;
Jie YANG
;
Feifei LIU
Author Information
1. 滨州医学院附属医院超声医学科,山东 滨州 256603
- Publication Type:Journal Article
- Keywords:
Breast neoplasms;
Ultrasonography;
Radiomics;
Peri-tumoral;
Lymphatic metastasis;
Pathology,surgical;
Forecasting
- From:
Chinese Journal of Medical Imaging
2025;33(10):1056-1062
- CountryChina
- Language:Chinese
-
Abstract:
Purpose To investigate the value of intratumoral and different ranges of peritumoral radiomics features of the primary lesion of breast cancer based on ultrasound images in predicting axillary lymph node metastasis(ALNM),and to explore the best peritumoral range.Materials and Methods A total of 312 cases confirmed by pathology in breast cancer patients with preoperative ultrasound images from June 2022 to February 2024 in Binzhou Medical University Hospital were retrospectively enrolled,and were randomly divided into training set and testing set according to the 7∶3 proportion.The tumor border of the ultrasound images was manually delineated as the intratumoral region of interest,and the peritumoral region of interest was obtained by conformal automatically extended different range(1,2,3,4 and 5 mm).The radiomics features were screened.Based on the selected optimal radiomics features,random forest classifier was used to construct three types of radiomics models(intratumoral model,5 peritumoral models,and 5 intratumoral+peritumoral models).The performance and clinical practicability of the models was assessed the area under the curve(AUC)and decision curve analysis.Results The AUCs of the intratumoral+peritumoral radiomics models for predicting ALNM in the training set and test set were 0.807-0.873,0.728-0.780,respectively,which were superior to those of the single intratumoral radiomics models(0.822,0.758)and peritumoral radiomics models(0.722-0.768,0.650-0.710).The intratumoral+peritumoral 3 mm radiomics model showed the best predictive performance,with AUC of 0.873 in the training set and 0.780 in the test set,respectively,and the decision curve showed that the model had a good clinical net benefit.Conclusion The combined intratumoral and peritumoral radiomics features of the primary lesion of breast cancer based on ultrasound images can effectively predict ALNM,and 3 mm peritumoral may be the best peritumoral range for predicting ALNM.
