Artificial Intelligence-Assisted Liver Ultrasound Training: Image Recognition and Real-Time Feedback System Based on
Ting Liu1, Qin Wen1, Jiaqi Yang2
1Department of Ultrasonography, The Fifth People's Hospital of Chengdu, Chengdu, China (T.L., Q.W., Y.W.).
Academic Radiology
|August 6, 2026
Summary
This study developed an AI-assisted liver ultrasound training system using YOLOv11n object detection and DeepSeek large language models. The AI system significantly improved training efficiency and learning outcomes compared to traditional methods.
Area of Science:
- Medical Education Technology
- Artificial Intelligence in Healthcare
- Medical Imaging Analysis
Background:
- Traditional liver ultrasound training faces challenges in efficiency and feedback.
- Developing advanced training tools is crucial for medical education.
Purpose of the Study:
- To develop and evaluate an AI-assisted liver ultrasound training system.
- To combine YOLOv11n object detection and DeepSeek large language models for enhanced ultrasound education.
Main Methods:
- A parallel controlled study with 20 students (10 traditional, 10 AI-assisted).
- YOLOv11n model trained on 326 liver ultrasound images.
- Comparison of system efficiency, accuracy, consistency, and post-training scores.
Main Results:
- YOLOv11n model achieved high performance (precision 0.984, recall 1.00, mAP@0.5 0.995).
- AI system reduced report processing time by 96.3% and feedback latency by 95.3%.
- Section recognition accuracy increased to 96.0%, evaluation consistency to 99.6%, and exam scores to 96.2.
Conclusions:
- AI-assisted training significantly enhances teaching efficiency and learning performance.
- Integration of computer vision and LLMs is effective for liver ultrasound education.
- The developed system shows promise for improving medical training outcomes.