跳到正文
arXiv cs.CL· Ruochen Li, Kun Yuan, Yufei Xia, Yue Zhou, Qingyu Lu, Weihang Li, Youxiang Zhu, Nassir Navab·· 2 天前AI 评分31

超越轨迹相似度:以协议依从性评估手术规划

Beyond Trajectory Similarity: Evaluating Surgical Planning by Protocol Adherence

AI 导读

研究提出以专家定义的协议规则为基础的手术规划评估框架,直接衡量协议依从性,而非与单条参考轨迹的相似度。结果显示,相似度指标会惩罚偏离参考轨迹但符合协议的规划,却奖励表面相似但违反协议的规划。对 Video-LLM 的评估还揭示两个可分离的失败模式:视频到状态 grounding 不可靠与规划约束不足,仅靠模型扩展无法解决,而注入结构化知识能带来稳定提升。

正文 · 原文

View PDF HTML (experimental)

Abstract:Surgical planning is commonly evaluated by trajectory similarity to a single observed execution, a paradigm borrowed from activity anticipation benchmarks such as Ego4D. This conflates two distinct concepts: task completion and protocol adherence and assumes similarity to a completed surgical trajectory implies a valid plan. In surgery, this assumption is flawed: the clinically relevant target is protocol adherence rather than mere completion, and the reference trajectory is only one observed execution that may itself deviate from the protocol. We therefore introduce an evaluation framework grounded in expert-defined protocol rules that directly measures protocol adherence. Under this framework, similarity-based metrics systematically fail: they penalize protocol-adherent plans that deviate from the reference trajectory, while rewarding protocol-violating plans that remain superficially similar to it. Evaluating Video-LLMs further reveals two separable failure modes: unreliable video-to-state grounding and under-constrained planning. Neither is resolved by model scaling alone, whereas structural knowledge injection provides consistent gains. Our results motivate a shift from trajectory similarity toward protocol adherence as the evaluation standard for surgical planning.
Subjects: Computation and Language (cs.CL); Robotics (cs.RO)
Cite as: arXiv:2601.10455 [cs.CL]
  (or arXiv:2601.10455v2 [cs.CL] for this version)
  https://doi.org/10.48550/arXiv.2601.10455

arXiv-issued DOI via DataCite

Submission history

From: Ruochen Li [view email]
[v1] Thu, 15 Jan 2026 14:47:26 UTC (918 KB)
[v2] Fri, 13 Nov 2026 12:39:43 UTC (2,566 KB)

来源:arXiv cs.CL · arxiv.org