多参数MRI影像组学术前预测乳腺癌HER-2表达状态的研究进展

Research Progress in Preoperative Prediction of Breast Cancer HER-2 Expression Status Using Multiparametric MRI-Based Radiomics

摘要: 乳腺癌作为全球女性发病率最高的恶性肿瘤，其精准诊疗对改善患者预后至关重要。人表皮生长因子受体2 (Human Epidermal Growth Factor Receptor-2, HER-2)表达状态是乳腺癌精准治疗方案选择的核心依据，传统检测依赖术后病理活检，存在有创性、取样误差及时间滞后等局限。多参数磁共振成像(Multiparametric Magnetic Resonance Imaging, mp-MRI)凭借多序列成像优势，可捕获肿瘤微观病理特征；影像组学技术则能将影像视觉信息转化为高通量量化特征，结合机器学习与深度学习算法，实现肿瘤生物学特征的无创性评估，为术前无创预测HER-2表达提供新路径。本文系统梳理2018~2025年mp-MRI影像组学在乳腺癌HER-2表达预测中的研究进展，从技术流程、序列选择、模型性能、临床价值等方面展开分析，指出当前研究存在的样本量局限、特征重复性不足、临床转化滞后等问题，并提出未来优化方向，为该技术的临床应用与研究深化提供参考。

Abstract: Breast cancer, as the malignant tumor with the highest incidence rate among women globally, its precise diagnosis and treatment are crucial for improving patient prognosis. The expression status of human epidermal growth factor receptor 2 (HER-2) is the core basis for selecting precise treatment options for breast cancer. Traditional detection relies on postoperative pathological biopsy, which has limitations such as invasiveness, sampling error, and time lag. Multi-parametric magnetic resonance imaging (mp-MRI), leveraging the advantages of multi-sequence imaging, can capture the microscopic pathological characteristics of tumors. Imaging omics technology can convert visual imaging information into high-throughput quantitative features. Combined with machine learning and deep learning algorithms, it enables noninvasive assessment of tumor biological characteristics, providing a new approach for preoperative noninvasive prediction of HER-2 expression. This article systematically reviews the research progress of mp-MRI radiomics in predicting HER-2 expression in breast cancer from 2018 to 2024. It analyzes the research from the perspectives of technical process, sequence selection, model performance, and clinical value. It points out the current issues in research, such as limited sample size, insufficient feature reproducibility, and lagging clinical translation. Furthermore, it proposes directions for future optimization, providing a reference for the clinical application and further research of this technology.