Related Experiment Video
Updated: Jun 26, 2026
15:48
Tracking the Mammary Architectural Features and Detecting Breast Cancer with Magnetic Resonance Diffusion Tensor Imaging
Published on: December 15, 2014
Combining Conventional MRI and DCE-MRI Radiomics for Prediction of Breast Cancer Molecular Subtypes: A Retrospective
Tran Thi Hue1,2, Nguyen Thu Huong2, Nguyen Duy Hung1,3
1Department of Radiology, Hanoi Medical University, Hanoi, Vietnam.
La Clinica Terapeutica
|June 24, 2026
Summary
Radiomic features from DCE-MRI significantly improve breast cancer molecular subtype prediction beyond conventional MRI. Combining radiomic and conventional MRI features offers the best diagnostic performance, especially for HER2 and triple-negative breast cancers.
Area of Science:
- Oncology
- Radiology
- Medical Imaging
Background:
- Accurate prediction of breast cancer molecular subtypes is crucial for guiding treatment decisions.
- Conventional MRI features offer some predictive capability but can be limited.
- Radiomics, the extraction of quantitative features from medical images, presents a novel approach to enhance diagnostic accuracy.
Purpose of the Study:
- To determine if radiomic features from dynamic contrast-enhanced MRI (DCE-MRI) improve breast cancer molecular subtype prediction compared to conventional MRI.
- To compare the predictive performance of models based on conventional MRI features, radiomics alone, and a combination of both.
Main Methods:
- A retrospective study included 206 primary breast cancer patients who underwent pretreatment 3.0-T breast MRI.
- Molecular subtypes were determined by immunohistochemistry and FISH/dual ISH.
- Radiomic features were extracted from segmented DCE-MRI images, harmonized, filtered, and selected using LASSO regression.
- Predictive models (MRI-based, radiomics-based, combined) were constructed using logistic regression.
Main Results:
- The combined model demonstrated superior performance across all subtypes: Luminal A (AUC=0.788), Luminal B (AUC=0.732), HER2-enriched (AUC=0.858), and triple-negative breast cancer (TNBC) (AUC=0.890).
- Radiomics alone showed improved performance over conventional MRI alone for some subtypes.
- The incremental value of radiomics was most pronounced for HER2-enriched and TNBC subtypes.
Conclusions:
- DCE-MRI radiomics provides significant added value for predicting breast cancer molecular subtypes beyond conventional MRI.
- Integrating radiomic and conventional MRI features yields the highest diagnostic discrimination, particularly beneficial for HER2-enriched and TNBC.
- This combined approach enhances personalized treatment strategies for breast cancer patients.
