中国首个全球冠军！宁诺方案教AI“看懂”手术，推动医疗机器人发展

Please scroll down for the English version - UNNC Team Wins Global Championship in AI-Powered Surgical Recognition.

近日，第29届国际医学图像计算和计算机辅助干预会议（MICCAI）公布了MICCAI 2026挑战赛各项赛事的获奖名单。宁波诺丁汉大学何祥健教授带领的学生团队在“手术动作三元组识别（MultiBypass Surgical Action Triplet Challenge 2026）”赛事中斩获全球第一名，系中国高校首次在该项赛事中夺冠。

何祥健教授（左）与获奖学生

近年来，宁诺在MICCAI挑战赛中表现亮眼，曾在2023年至2025年间荣膺“国际脑肿瘤图像修复”赛事全球三连冠。

作为全球医学影像计算领域的重量级赛事，今年的MICCAI挑战赛吸引了牛津大学、耶鲁大学、复旦大学、德国癌症中心等千余所全球知名高校和机构参与角逐。

先看懂、再动手

宁诺方案助推医疗机器人发展

“当前，手术录像在大多数医院仍以回顾、教学、存档等用途为主。”博士生徐卿是本次获奖队伍的队长，他介绍道：“比赛中，我们搭建了一套分析手术实时画面的AI模型。通过识别手术动作三元组，即‘器械（什么工具）、动作（做了什么）和目标（作用对象）’，这套模型能够实时解读医生的每个步骤——手术还在进行，AI就能同步识别，为手术的实时监测与智能化分析提供支撑。”

据徐卿介绍，团队创新运用因果时序残差等技术手段，不仅显著提高了识别手术动作三元组的准确率，还使模型具备更强的泛化能力，能够灵活适配不同医疗机构的多种场景。

宁诺学生白欣儒（左）与徐卿（右）

实际上，这份竞赛方案的意义远不止于夺冠。团队成员、大四学生白欣儒阐述了其应用前景。

“随着医学影像计算的发展，机器人将在未来走进手术室并承担起操作任务。”他解释道，“为了实现这一点，机器人首先需要具备分析复杂医疗场景、推断合理操作、研判潜在风险的能力。可以说，机器人对手术动作三元组的识别和分析是其迈向自主操作的关键基础。”

调资源、广协同

名师助力团队成功登顶

在团队的参赛过程中，指导老师何祥健教授始终扮演着"领路人"的角色。作为宁诺计算机科学与技术首席教授，他长期深耕计算机视觉、图像处理与行为识别领域，多年入选"全球前2%顶尖科学家"榜单。这样一位"学术大牛"带团队，却很舍得"放手"——大方向他来把关，具体怎么走交给学生去闯。

“备赛期间，识别准确率无法提升曾一度困扰着团队的开发进程。”徐卿回忆，“何教授没有直接告诉我们怎么改，而是带我们重新审视模型整体搭建，点出可以优化的方向，随后便让我们自行规划优化方案。这让团队有了更多发挥和探索的空间，整体思路也能保持更好的延续性。”

除了技术指导，何祥健教授还为团队拉来了香港理工大学及其“朋友圈”。通过接触临床一线人士和业内专家，团队得以在真实的医疗场景中不断迭代模型，让算法真正“读懂”手术逻辑，提升模型的准确率和泛化能力。

目前，团队正基于本次获奖模型撰写学术论文，计划投递国际顶级期刊。

UNNC Team Wins Global Championship in AI-Powered Surgical Recognition

A student team led by Professor Xiangjian He from the University of Nottingham Ningbo China (UNNC) has won first place globally in the MultiBypass Surgical Action Triplet Challenge 2026, marking the first time a Chinese university has claimed the title.

The team developed an AI model that can recognise surgical instruments, actions and targets in real time, enabling it to interpret surgical procedures as they unfold. The technology could support more intelligent surgical monitoring and analysis, laying important groundwork for the future development of autonomous medical robots.

The team improved the model’s recognition accuracy and ability to adapt to different clinical settings through innovative technical approaches. Collaboration with clinical professionals and industry experts, facilitated by Professor He, also allowed the students to refine their model using real-world medical scenarios, helping bridge the gap between algorithm development and practical healthcare applications.

Professor He, Chair Professor of Computer Science and Technology at UNNC, guided the team by providing strategic direction while encouraging students to explore their own solutions. This approach gave the team greater scope for experimentation and independent problem-solving, demonstrating the university’s commitment to fostering student innovation through research-led learning and expert mentorship.

The achievement builds on UNNC’s strong track record in the MICCAI challenges, following three consecutive global titles in the International Brain Tumour Image Segmentation challenge from 2023 to 2025. This latest success further demonstrates the university’s growing contribution to medical image computing and its efforts to empower students to tackle complex global challenges through interdisciplinary research and innovation.

The team is now preparing an academic paper based on its award-winning model for submission to a leading international journal.

图文来源：Patrick Cai，Tony Bao

电子编辑：Suri An

精彩内容

全球第97位！诺丁汉大学2027QS世界大学排名，稳了！

宁波诺丁汉大学正式获批推免资格，系首批拥有此资格的中外合作高校！

宁诺24/25就业报告：近50%深造学子获世界前10院校录取

宁波诺丁汉大学精彩视频

关注更多

宁波诺丁汉大学官方平台