Prediction of Lymphovascular Invasion in cN0 Breast Cancer Based on Multi-Parametric MRI Radiomics Features
10.3969/j.issn.1005-5185.2025.10.003
- VernacularTitle:基于多参数MRI影像组学特征预测临床淋巴结阴性乳腺癌淋巴血管浸润
- Author:
Shunian LI
1
;
Yiyan SHANG
;
Yaxin GUO
;
Jun LIAO
;
Yunxia WANG
;
Xiaodong LI
;
Meiyun WANG
;
Hongna TAN
Author Information
1. 郑州大学人民医院放射科,河南 郑州 450003
- Publication Type:Journal Article
- Keywords:
Breast neoplasms;
Lymphovascular invasion;
Magnetic resonance imaging;
Diffusion weighted imaging;
Clinically lymph node-negative;
Radiomics;
Forecasting
- From:
Chinese Journal of Medical Imaging
2025;33(10):1035-1042
- CountryChina
- Language:Chinese
-
Abstract:
Purpose To investigate the value of intratumoral and peritumoral radiomics features based on multi-parametric MRI for preoperative prediction of lymphovascular invasion(LVI)in clinical lymph node-negative(cN0)breast cancer.Materials and Methods This retrospective study included 280 patients with pathologically confirmed breast cancer who underwent preoperative MRI at Henan Provincial People's Hospital from January 2017 to May 2021.Patients were randomly divided into a training cohort and a testing cohort.After Z-score normalization,feature selection was performed using Select K Best and least absolute shrinkage and selection operator regression.Random forest algorithms were used to construct intratumoral,peritumoral,and combined intratumoral-peritumoral radiomics models for LVI prediction.Model performance and clinical utility were evaluated using the area under the receiver operating characteristic curve(AUC),calibration curves and decision curve analysis.Results High Ki-67 expression(≥20%),axillary lymph node metastasis and positive diffusion weighted imaging(DWI)margin sign were more common in the LVI-positive group(χ2=5.959,18.316,20.554,all P<0.05).In the testing cohort,the AUC values of the dynamic contrast-enhanced(DCE)-Intra and DCE-Com models for predicting LVI status were higher than those of the DWI sequence,whereas the AUC value of the DWI-Peri model was higher than that of the DCE sequence.The DWI-DCE-Com model achieved AUCs of 0.836 and 0.818 in the training and testing cohorts,respectively,which surpassed the predictive performance of single-sequence intratumoral-peritumoral radiomics models(DWI-Com,DCE-Com).Decision curve analysis showed that the DWI-DCE-Com model provided greater net clinical benefit across a reasonable range of threshold probabilities.Conclusion Radiomics models based on multiparametric MRI features from intratumoral and peritumoral regions can effectively predict LVI status in cN0 breast cancer,offering valuable support for preoperative individualized treatment decision-making.
