Abstract:

With the rising incidence of breast cancer， early diagnosis and accurate prognostic evaluation have become pivotal for improving survival rates and optimizing personalized treatment. Multimodal ultrasound， by integrating technologies such as elastography， viscoelastic imaging， contrast-enhanced ultrasound， microflow imaging， and photoacoustic/ultrasound， can provide multi-dimensional information regarding lesion morphology， mechanical properties， blood perfusion， and tissue function in a non-invasive， dynamic， and real-time manner. This compensates for the limitations of conventional grayscale ultrasound in characterizing microvasculature， tissue stiffness， and biomechanical properties. Artificial Intelligence （AI）， particularly deep learning methods， has demonstrated remarkable capabilities in automatic feature extraction from images， multimodal data fusion， and predictive modeling， thereby providing technical support for transforming ultrasound imaging from image acquisition into clinical decision-making support. This paper systematically reviews recent advances in multimodal ultrasound technologies and their integration with AI in the areas of early breast cancer diagnosis， benign-malignant differentiation， prediction of lymph node metastasis， and prediction of molecular subtypes， summarizing both research achievements and existing limitations. Furthermore， it explores the application value of AI in lesion detection， automatic segmentation， radiomics feature fusion， and the construction of prognostic prediction models based on multi-source clinical and imaging data. Key challenges such as model interpretability， data standardization， cross-institution generalization capability， and clinical acceptability are analyzed， and strategic recommendations for promoting clinical translation are proposed. The synergistic development of multimodal ultrasound and AI holds promise for evolving ultrasound from an auxiliary detection tool into a pivotal component of early diagnosis， treatment， and comprehensive management throughout the breast cancer care continuum. However， its widespread application still requires concerted advancement in technological maturity， clinical validation， and policy support.

Key words: multimodal ultrasound, breast cancer, artificial intelligence, viscoelastic imaging, super-resolution ultrasound imaging