# Machine Learning Model Based on Multiparametric MRI for Distinguishing HER2 Expression Level in Breast Cancer

## TL;DR

This study uses MRI-based machine learning to noninvasively classify HER2 expression levels in breast cancer, potentially aiding treatment decisions and survival prediction.

## Contribution

The novel contribution is the development of interpretable machine learning models using conventional MRI features to classify HER2 expression levels and their association with survival outcomes.

## Key findings

- Machine learning models achieved AUCs of 0.75 and 0.73 for distinguishing HER2-positive vs. HER2-negative tumors in internal and external test sets.
- Higher model scores in Task 1 were associated with shorter disease-free survival (p = 0.037).
- Key MRI features like tumor size and peritumoral edema were identified as important for HER2 classification.

## Abstract

Human epidermal growth factor receptor 2 (HER2) plays a crucial role in guiding treatment decisions and predicting outcomes in breast cancer. Recent advances indicate that patients with low HER2 expression may also benefit from novel targeted therapies, underscoring the need for accurate and noninvasive assessment of HER2 status. In this study, we developed machine learning models based on conventional magnetic resonance imaging (MRI) to classify different HER2 expression levels in invasive breast cancer. The models showed potential in distinguishing HER2-positive, HER2-low, and HER2-zero tumors and showed exploratory associations between model scores and survival outcomes. Using an interpretable artificial intelligence approach, we identified key MRI features related to HER2 status, such as tumor size, lymph node involvement, and peritumoral edema. These findings suggest that MRI-based machine learning may offer a noninvasive tool to support treatment planning, with possible implications for individualized risk stratification.

This study aimed to develop machine learning models based on conventional MRI features to classify HER2 expression levels in invasive breast cancer and explore their association with disease-free survival (DFS). A total of 678 patients from two centers were included, with Center 1 divided into training and internal test sets and Center 2 serving as an external test set. Random Forest models were trained to distinguish HER2-positive vs. HER2-negative (Task 1) and HER2-low vs. HER2-zero tumors (Task 2) using BI-RADS–based MRI features. SHapley Additive exPlanations were applied to rank feature importance, assist feature selection, and enhance model interpretability. DFS was analyzed using Kaplan–Meier curves and log-rank tests. In Task 1, key features included tumor size, axillary lymph nodes, fibroglandular tissue, peritumoral edema, and multifocal, achieving AUCs of 0.75 and 0.73 in the internal and external test sets, respectively. In Task 2, tumor size, peritumoral edema, and multifocal yielded AUCs of 0.73 and 0.72, respectively. Higher task-specific model scores were associated with shorter DFS in Task 1 (p = 0.037) and longer DFS in Task 2 (p = 0.046). MRI-based machine learning models can noninvasively stratify HER2 expression levels, with potential for prognostic stratification and clinical application.