上海口腔医学

上海口腔医学 ›› 2026, Vol. 35 ›› Issue (4): 440-444.doi: 10.19439/j.sjos.2026.04.015

• 医学教育 • 上一篇    下一篇

淋巴结人工智能辅助诊断系统在头颈影像教学中的应用

杨功鑫1, 戴晓庆1, 王晶波1, 巫智涵2, 陶晓峰1, 朱凌1   

  1. 1.上海交通大学医学院附属第九人民医院 放射科,上海 200011;
    2.上海交通大学医学院,上海 200025
  • 收稿日期:2025-01-07 修回日期:2025-03-22 出版日期:2026-08-25 发布日期:2026-09-01
  • 通讯作者: 朱凌,E-mail：puxuke12@126.com
  • 作者简介:杨功鑫（1986—）,男,博士,主治医师,E-mail：yanggongxin1986@163.com
  • 基金资助:
    2022国家级 “大创” 项目及“大学生创新训练计划”“医+X”交叉学科项目（1521X3）; 2019年教学型医师-朱凌第三期（JXXYS-2019-5）; 上海交通大学医学院附属第九人民医院“交叉”基金（JYJC202404）

Application of an artificial intelligence-assisted diagnostic system for lymph nodes in head and neck imaging teaching

Yang Gongxin1, Dai Xiaoqing1, Wang Jingbo1, Wu Zhihan2, Tao Xiaofeng1, Zhu Ling1   

  1. 1. Department of Radiology, Shanghai Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine. Shanghai 200011;
    2. Shanghai Jiao Tong University School of Medicine. Shanghai 200025, China
  • Received:2025-01-07 Revised:2025-03-22 Online:2026-08-25 Published:2026-09-01

PDF (PC)

11

摘要： 目的: 探讨基于CT影像的颈部淋巴结人工智能（artificial intelligence,AI）辅助诊断系统在口腔颌面头颈影像淋巴结识别教学中的可行性和效果。方法: 选择上海交通大学口腔医学院33名学生为研究对象,随机分为A、B两组,分别对两例颌面部增强CT影像进行阅片并标注淋巴结。对其中一例A组采用人工标注,B组在AI辅助下标注,随后对另一例两组互换标注方式进行交叉标注。以高年资医师的标注结果作为金标准,比较人工标注组与AI辅助标注组的召回率和准确度,并通过教学质量问卷评估学生的学习体验。结果: 人工标注组的平均召回率为31.0%,而在AI系统辅助下标注的召回率提升至63.9%。所有学生均完成教学问卷,其中93.9%的学生认为AI辅助诊断系统能够激发学习兴趣、加深对理论知识的理解,有助于头颈部影像解剖的学习。结论: 颈部淋巴结AI辅助诊断系统在口腔颌面头颈影像教学中具有显著优势,能够有效提高医学生对头颈影像解剖的理解和学习效果,具有较高的应用价值。

关键词: 人工智能辅助诊断, 颈部淋巴结, 影像教学

Abstract: PURPOSE: To explore the feasibility and effectiveness of a CT-based artificial intelligence (AI) assisted diagnostic system for cervical lymph node recognition in oral and maxillofacial head and neck imaging education. METHODS: A total of 33 students from College of Stomatology, Shanghai Jiao Tong University were selected as study participants. The students were randomly divided into group A and B, and tasked with reviewing the anatomy of cervical lymph nodes in contrast-enhanced CT images of two patients. Group A performed manual annotation, while group B used AI-assisted annotation, and then the groups crossed over to annotate the next patient. The annotation results of senior radiologists were used as the gold standard to compare the recall rate and accuracy between the manual annotation group and the AI-assisted annotation group. Additionally, a teaching quality questionnaire was used to assess students' learning experiences, focusing on the effectiveness of AI in enhancing engagement and comprehension. RESULTS: Statistical analysis showed that the average recall rate for manual annotation was 31.0%, whereas the recall rate increased to 63.9% with AI assistance. All students completed the teaching questionnaire, and 93.9% of them believed that the AI-assisted diagnostic system stimulated their interest in learning, deepened their understanding of theoretical knowledge, and facilitated their study of head and neck imaging anatomy. Furthermore, students reported that AI assistance helped in reducing cognitive load, allowing them to focus more on understanding the complex anatomical relationships rather than manual tasks. CONCLUSIONS: The AI-assisted diagnostic system for cervical lymph nodes has significant educational advantages in oral and maxillofacial head and neck imaging teaching. It effectively improves medical students' understanding of head and neck imaging anatomy, reduces cognitive load, and enhances learning outcomes. This study demonstrates the high application value of integrating AI into radiology education, providing a model for future educational innovations and reforms.

Key words: AI-assisted diagnosis, Cervical lymph nodes, Imaging teaching

中图分类号: