BACKGROUND AND OBJECTIVE

As the detection rate of thyroid nodules increases year by year, traditional ultrasonic diagnostic methods face challenges such as inefficiency and high dependence on physician experience. This paper focuses on the research status, advantages and challenges of robot automatic scanning and intelligent diagnosis system.

METHODS

We systematically retrieved the PubMed and Web of Science databases, screened and integrated relevant articles, and conducted a systematic analysis and summary of the existing research.

KEY CONTENT AND FINDINGS

The development of robot and artificial intelligence (AI) provides a new method for efficient and accurate ultrasound diagnosis of thyroid nodules. Robot enables automated scanning of thyroid through precise robotic arm control, positioning, and trajectory planning, significantly improving the standardization and repeatability of the diagnostic process. However, its flexibility in clinical application and patient acceptance still needs to be further improved. From the early rule matching research based on manual features to the automatic feature processing of thyroid nodules using deep learning algorithms have made AI outstanding in the ultrasound diagnosis of thyroid nodules. Meanwhile, the innovative research of deep learning in the contrast-enhanced ultrasound (CEUS) video analysis has broadened the application of intelligent diagnosis systems. The interpretability of the deep learning models is solved to some extent by Gradient-weighted Class Activation Mapping (Grad-CAM) and other techniques. However, the interpretability, data dependence, and ability to generalize deep learning models in clinical practice remain key issues to be addressed.

CONCLUSIONS

Robots and AI have brought revolutionary progress to the diagnosis of thyroid diseases, but their clinical translational application still faces many challenges.