Contrast-enhanced mammography-based interpretable machine learning model for the prediction of the molecular subtype breast cancers

Sep 23, 2026

Uploaded by Aditya

Objective This study establishes a machine learning prediction model linking contrast-enhanced mammography (CEM) imaging features to molecular subtypes of mass-type breast cancer. Materials and Methods A retrospective cohort of 241 women underwent preoperative CEM between 2018 and 2021, split 7:3 into training and test sets. Twenty-one visually described features (four clinical, seventeen radiological) were extracted, followed by binary subtype tasks and classification using multinomial naive Bayes with LASSO-based feature selection. SHAP was used for interpretability. Results Using low-energy and dual-energy subtraction (LE+DES) images achieved the best ROC AUC across subtype comparisons, with key SHAP features driving Luminal/non-Luminal predictions. Conclusions CEM-derived radiological characteristics are associated with breast cancer molecular subtypes and require external validation.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages11

Reading Time28 min

AccessFree to download

Abstract

Objective

Materials and Methods

Results

Conclusions

Keywords

Introduction

Ma etal. BMC Medical Imaging (2025) 25:255 [https://doi.org/10.1186/s12880-025-01765-3](https://doi.org/10.1186/s12880-025-01765-3) BMC Medical Imaging RESEARCH Open Access Contrast-enhanced mammography-based interpretable machine learning model for the prediction of the molecular subtype breast cancers Mengwei Ma 1†, Weimin Xu 1†, Jun Yang 1, Bowen Zheng 1, Chanjuan Wen 1, Sina Wang 1, Zeyuan Xu 1, Genggeng Qin 1* and Weiguo Chen 1* Abstract Objective This study aims to establish a machine learning prediction model to explore the correlation between contrast-enhanced mammography (CEM) imaging features and molecular subtypes of mass-type breast cancer. Materials and Methods This retrospective study included women with breast cancer who underwent CEM preoperatively between 2018 and 2021. We included 241 patients, which were randomly assigned to either a training or a test set in a 7:3 ratio. Twenty-one features were visually described, including four clinical features and seventeen radiological features, these radiological features which extracted from the CEM. Three binary classifications of subtypes were performed: Luminal vs. non-Luminal, HER2-enriched vs. non-HER2-enriched, and triple-negative (TNBC) vs. non-triple-negative. A multinomial naive Bayes (MNB) machine learning scheme was employed for the classification, and the least absolute shrink age and selection operator method were used to select the most predictive features for the classifiers. The classification performance was evaluated using the area under the receiver operating characteristic curve. We also utilized SHapley Additive exPlanation (SHAP) values to explain the prediction model. Results The model that used a combination of low energy (LE) and dual-energy subtraction (DES) achieved the best performance compared to using either of the two images alone, yielding an area under the receiver operating characteristic curve of 0.798 for Luminal vs. non-Luminal subtypes, 0.695 forTNBC vs. non-TNBC, and 0.773 for HER2-enriched vs. non-HER2-enriched. The SHAP algorithm shows that“LE_mass_margin_spiculated,”“DES_mass_enhanced_margin_spiculated,” and“DES_mass_internal_enhancement_homogeneous” have the most significant impact on the model’s performance in predicting Luminal and non-Luminal breast cancer.“mass_calcification_ †Mengwei Ma and Weimin Xu contributed equally to this work. *Correspondence: Genggeng Qin [zealotq@smu.edu.cn](zealotq@smu.edu.cn)[ ](zealotq@smu.edu.cn)Weiguo Chen[chen1999@smu.edu.cn](chen1999@smu.edu.cn) Full list of author information is available at the end of the article © The Author(s) 2025. Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit [http://creati](http://creati)[vecommons.org/l](vecommons.org/l)icenses/by-nc-nd/4.0/. Ma et al. BMC Medical Imaging (2025) 25:255 Page 2 of 11 relationship_no,”“calcification_ type_no,” and“LE_mass_margin_spiculated” have a considerable impact on the model’s performance in predicting HER2 and non-HER2 breast cancer. Conclusions The radiological characteristics of breast tu

FAQs

What is the main objective of the study?

To build a machine learning prediction model that relates contrast-enhanced mammography (CEM) imaging features to the molecular subtypes of mass-type breast cancer.

How were participants selected and how was the dataset split?

Women with breast cancer who underwent preoperative CEM between 2018 and 2021 were included, totaling 241 patients. The dataset was randomly assigned to a training and a test set in a 7:3 ratio.

What model and feature selection approach were used?

A multinomial naive Bayes scheme performed the classification, while least absolute shrinkage and selection operator (LASSO) was used to select the most predictive features for each classifier.

Sep 23, 2026

Uploaded by Aditya