Correlation between Automated Breast Volume Scanner Features and Molecular Typing in Breast Cancer: A Literature Review

乳腺癌的分子分型对指导临床治疗决策和预后评估具有重要意义。传统病理活检作为分子分型诊断的金标准，存在侵入性操作、肿瘤异质性及时效性差等局限。近年来，自动乳腺全容积成像(Automated Breast Volume Scanner, ABUS/ABVS)技术，特别是其冠状面成像特征，在无创预测乳腺癌分子分型方面展现出重要价值。本文通过系统回顾相关文献，深入探讨ABUS/ABVS的影像学特征与乳腺癌分子分型(Luminal A型、Luminal B型、HER2过表达型及三阴型)之间的关联机制，分析其在分化程度评估、预后预测中的效能，并综述放射组学模型与多模态融合技术的前沿进展。研究表明，不同分子分型在ABUS冠状面上呈现显著差异：“汇聚征”与Luminal型密切相关，微钙化在HER2过表达型中高发，而三阴型多表现为边界清晰和形态规则。基于ABUS的放射组学模型进一步将分子分型预测AUC提升至0.82~0.87，为精准医疗提供了新方向。本综述还讨论了当前技术临床转化面临的挑战，包括操作标准化、样本异质性等问题，并对深度学习与动态功能成像等未来发展方向提出展望。Molecular typing of breast cancer is of great significance in guiding clinical treatment decisions and prognostic evaluation. Traditional pathological biopsy, as the gold standard for molecular typing diagnosis, has limitations such as invasive procedures, tumor heterogeneity and poor timeliness. In recent years, Automated Breast Volume Scanner (ABUS/ABVS) technology, especially its coronal imaging features, has shown great value in non-invasive prediction of breast cancer molecular typing. In this paper, we systematically review the relevant literature to explore the association mechanism between the imaging characteristics of ABUS/ABVS and the molecular typing of breast cancer (Luminal A, Luminal B, HER2 overexpression and triple negative type), analyze its efficacy in the evaluation of differentiation degree and prognosis, and review the cutting-edge progress of radiomics models and multimodal fusion technology. The results showed that different molecular subtypes showed significant differences in the ABUS coronal plane: the “convergence sign” was closely related to the Luminal type, and microcalcifications were more common in the HER2 over-expression type, while the triple-negative type was mostly characterized by clear boundaries and regular morphology. The ABUS-based radiomics model further improved the predicted AUC of molecular typing to 0.82~0.87, providing a new direction for precision medicine. This review also discusses the current challenges in the clinical translation of technologies, including operational standardization and sample heterogeneity, and puts forward prospects for future development directions such as deep learning and dynamic functional imaging.