Machine learning models based on magnetic resonance imaging for predicting Lymphovascular Invasion in Invasive Breast Cancer
PLOS One · 29.05.2026 · 21 (5) · e0350085
DOI: 10.1371/journal.pone.0350085 ↗Kurz gefasst
Fachartikel in der Zeitschrift PLOS One, online erschienen am 29.05.2026. Verfasst von Hong Li und 5 weiteren Autor:innen. Beteiligte Einrichtung: Department of Radiology, the Second Affiliated Hospital, Guangzhou Medical University, Guangzhou, Guangdong, China. Hauptthemen laut MEDLINE: Breast Neoplasms, Machine Learning, Magnetic Resonance Imaging, Lymphatic Metastasis. Die Arbeit nennt 33 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Objectives
Treatment strategies for invasive breast cancer require accurate lymphovascular invasion (LVI) predictions. This study aimed to investigate the feasibility and effectiveness of delta radiomics signature based on dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) and radiomics signature based on T2-weighted fat suppressed imaging(T2FS) for assessing LVI in invasive breast cancer. Materials and methods A total of 166 patients with resectable invasive breast cancer who underwent preoperative DCE-MRI and T2FS from July 10, 2020 to December 31, 2023 were enrolled. Radiomics features were extracted from pre-contrast phase (RF pre ), second post-contrast phase (RF post ), and Delta radiomics features (RF Delta ) calculated as RF post minus RF pre . Then four radiomics signatures (RS T2 , RS pre , RS post , RS Delta ) were further developed based on the Random Forest model for RF T2 , RF pre , RF post and RF Delta , respectively. The predictive performance of all signatures was evaluated by receiver operating characteristic (ROC) analysis, with accuracy and area under the curve (AUC) as the main quantitative metrics.
Results
In the test set, RS Delta (10 features) achieved the highest accuracy of 0.717 and an AUC of 0.764; RS post (8 features) had an accuracy of 0.565 and an AUC of 0.610; RS pre (7 features) and RS T2 (6 features) both showed an accuracy of 0.565 with AUCs of 0.535 and 0.662, respectively. Statistical differences were observed in predictive performance between RS Delta and RS pre , RS post (both p < 0.05), while no significant difference was found between RS Delta and RS T2 ( p = 0.239).
Conclusion
RS T2 , RS pre , RS post and RS Delta are all feasible for predicting LVI in invasive breast cancer, and RS Delta outperforms the other three radiomics signatures, which can serve as a potential non-invasive imaging biomarker for LVI prediction in clinical practice.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Autor:innen und Institutionen
- Department of Radiology, the Second Affiliated Hospital, Guangzhou Medical University, Guangzhou, Guangdong, China.
- Department of Radiology, Guangzhou Geriatric Hospital, Guangzhou, Guangdong, China.
- Department of Radiology, Guangzhou Women And Children's Medical Center, Guangzhou, China.
- Department of Breast surgery, the Second Affiliated Hospital, Guangzhou Medical University, Guangzhou, Guangdong, China.
- Department of Pathology, the Second Affiliated Hospital, Guangzhou Medical University, Guangzhou, Guangdong, China.
ORCID-Profile: Hong Li
Zitieren
APA
Li, H., Huang, J., Hou, J., Chen, X., Zhi, C., & Li, Z. (2026). Machine learning models based on magnetic resonance imaging for predicting Lymphovascular Invasion in Invasive Breast Cancer. PLOS One, 21(5), e0350085. https://doi.org/10.1371/journal.pone.0350085
Vancouver
Li H, Huang J, Hou J, Chen X, Zhi C, Li Z. Machine learning models based on magnetic resonance imaging for predicting Lymphovascular Invasion in Invasive Breast Cancer. PLOS One. 2026;21(5):e0350085. doi:10.1371/journal.pone.0350085
BibTeX
@article{Li2026,
author = {Li, Hong and Huang, Jieling and Hou, Jianning and Chen, Xinxin and Zhi, Cheng and Li, Zhiming},
title = {Machine learning models based on magnetic resonance imaging for predicting Lymphovascular Invasion in Invasive Breast Cancer},
journal = {PLOS One},
year = {2026},
volume = {21},
number = {5},
pages = {e0350085},
doi = {10.1371/journal.pone.0350085}
}
EUVIMED