Abstract: Ultrasonography is the primary imaging modality for evaluating thyroid nodules and cervical lymph nodes.However,traditional diagnostic approaches heavily rely on physician expertise,presenting limitations such as inconsistent results and operator-dependent variability.The application of artificial intelligence(AI),particularly deep learning,in thyroid and cervical lymph node ultrasonography is advancing the field toward automation and standardization.AI-assisted diagnostic systems demonstrate remarkable performance in differentiating benign from malignant thyroid nodules,achieving accuracy comparable to or surpassing that of senior physicians,with some products already implemented in clinical practice.Nevertheless,diagnostic accuracy for cervical lymph node metastasis requires further improvement,and dedicated products remain under development.Additionally,the clinical translation of AI faces challenges including data limitations,insufficient model interpretability,suboptimal human-computer collaboration,and regulatory-ethical concerns.Future directions will focus on model optimization,multimodal integration,and enhanced human-AI collaboration to drive comprehensive advancements in diagnostic and therapeutic standards.

Key words: artificial intelligence; thyroid nodules; cervical lymph nodes; ultrasonographic diagnosis