Surg-R1: A Hierarchical Reasoning Foundation Model for Scalable and Interpretable Surgical Decision Support with Multi-Center Clinical Validation

TL;DR

Surg-R1 is a hierarchical surgical vision-language model with a large reasoning dataset, trained via a multi-stage pipeline, achieving superior interpretability and performance across multiple surgical understanding benchmarks.

Contribution

Introduces a three-level hierarchical reasoning framework, the largest surgical reasoning dataset, and a four-stage training pipeline for scalable and interpretable surgical decision support.

Findings

Achieves highest Arena Score of 64.9% on public benchmarks.

Outperforms existing models on multiple surgical tasks.

Improves external validation performance by 15.2 percentage points.

Abstract

Surgical scene understanding demands not only accurate predictions but also interpretable reasoning that surgeons can verify against clinical expertise. However, existing surgical vision-language models generate predictions without reasoning chains, and general-purpose reasoning models fail on compositional surgical tasks without domain-specific knowledge. We present Surg-R1, a surgical Vision-Language Model that addresses this gap through hierarchical reasoning trained via a four-stage pipeline. Our approach introduces three key contributions: (1) a three-level reasoning hierarchy decomposing surgical interpretation into perceptual grounding, relational understanding, and contextual reasoning; (2) the largest surgical chain-of-thought dataset with 320,000 reasoning pairs; and (3) a four-stage training pipeline progressing from supervised fine-tuning to group relative policy optimization and iterative self-improvement. Evaluation on SurgBench, comprising six public benchmarks and six multi-center external validation datasets from five institutions, demonstrates that Surg-R1 achieves the highest Arena Score (64.9%) on public benchmarks versus Gemini 3.0 Pro (46.1%) and GPT-5.1 (37.9%), outperforming both proprietary reasoning models and specialized surgical VLMs on the majority of tasks spanning instrument localization, triplet recognition, phase recognition, action recognition, and critical view of safety assessment, with a 15.2 percentage point improvement over the strongest surgical baseline on external validation.