Numerous research studies have revealed that breast cancer is characterized by substantial biological and molecular heterogeneity, which can neither be fully captured by conventional clinical medical imaging nor biopsy which is typically limited by sampling error and invasiveness. Despite the exhibited complexity, developments in ultrasound imaging technologies from morphological imaging only to multi-dimensional functional imaging coupled with quantitative analysis have significantly improved the diagnostic accessibility and efficiency for breast cancer patients. This review first summarizes recent advances in multi-modal ultrasound imaging encompassing microvascular and super-resolution ultrasound imaging with contrast enhancement, elastography and viscoelastic imaging, as well as quantitative analysis consisting of radiofrequency signal analysis, radiomics, and artificial intelligence-based image analysis for precision breast cancer diagnosis. Specific attention is given to their applications in benign-malignant differentiation, molecular subtype characterization, lymph node metastasis assessment. Beyond image-based analysis, the integration of ultrasound radiomics with genomic, transcriptomic, and other multi-omics data is emerging as a promising strategy for non-invasive molecular phenotyping, although its clinical application is still at its infancy. This review then discusses the transition from radiomics alone through radiogenomics to radio-multiomics, highlighting the potential of advanced ultrasound imaging technologies to bridge imaging phenotypes with underlying tumor biological behaviour in order to facilitate individualized diagnosis and treatment planning. Most radio-multiomics approaches remain at a preclinical or proof-of-concept stage and still require validation in large-scale paired cohorts. In addition, issues such as standardization, reproducibility, interpretability and data heterogeneity remain to be resolved. Finally, this review discusses current challenges that should be addressed before ultrasound-based radio-multi-omics can be translated to a viable clinical tool.

Accurate preoperative prediction of cervical Lymph Node Metastasis (LNM) is critical for surgical decision-making in thyroid cancer patients, and the difficulty in it often leads to over-treatment. UltraSound (US) and Computed Tomography (CT) are two primary non-invasive examinations, but neither method alone provides satisfactory diagnostic accuracy. To address this problem, we propose a Multimodal Nested Attention Network (MNANet) to integrate US and CT images. The network is designed to extract specific complementary information from US and CT images, and comprehensively fuse multimodal features at multiple granularities. In our internal cohort, MNANet achieves Areas Under the Curves (AUCs) of 0.88 and 0.86 for central and lateral cervical sites, respectively, representing a significant improvement of 0.06 to 0.10 compared to unimodal models and outperforming state-of-the-art medical multimodal methods across all metrics.The model demonstrates robust cross-institutional generalization and maintains superior performance across other imaging modalities(e.g., Magnetic Resonance Imaging (MRI)). Additionally, our model exhibits a more precise focus on the thyroid nodule, indicating enhanced learning ability. Moreover, we systematically evaluate the applicability across various clinical characteristics, identifying individuals who can benefit most from the multimodal approach.