PhySe-RPO: Physics and Semantics Guided Relative Policy Optimization for Diffusion-Based Surgical Smoke Removal

2026-03-24

AI Summary

This work addresses the critical challenge of surgical smoke obscuring intraoperative anatomy and degrading visual perception, a problem inadequately tackled by existing desmoking methods that rely on scarce paired supervision and lack exploratory capability. To overcome these limitations, the study introduces relative policy optimization into a diffusion-based desmoking framework, enabling reference-free stochastic restoration through physical consistency constraints and CLIP-driven semantic rewards. This novel approach facilitates trajectory-level exploration and eliminates the need for a critic in policy updates. Evaluated on both synthetic and real robotic surgery datasets, the method produces desmoked results that are physically plausible, semantically accurate, and clinically interpretable, significantly outperforming current state-of-the-art techniques and enhancing the robustness and practicality of surgical video restoration.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Search and Optimization: Learning to Search
• Intelligent Robots: Learning & Optimization for ROB

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Surgical smoke severely degrades intraoperative video quality, obscuring anatomical structures and limiting surgical perception. Existing learning-based desmoking approaches rely on scarce paired supervision and deterministic restoration pipelines, making it difficult to perform exploration or reinforcement-driven refinement under real surgical conditions. We propose PhySe-RPO, a diffusion restoration framework optimized through Physics- and Semantics-Guided Relative Policy Optimization. The core idea is to transform deterministic restoration into a stochastic policy, enabling trajectory-level exploration and critic-free updates via group-relative optimization. A physics-guided reward imposes illumination and color consistency, while a visual-concept semantic reward learned from CLIP-based surgical concepts promotes smoke-free and anatomically coherent restoration. Together with a reference-free perceptual constraint, PhySe-RPO produces results that are physically consistent, semantically faithful, and clinically interpretable across synthetic and real robotic surgical datasets, providing a principled route to robust diffusion-based restoration under limited paired supervision.

Problem

• surgical smoke removal
• intraoperative video quality
• paired supervision
• deterministic restoration
• visual perception

Innovation

• diffusion-based restoration
• relative policy optimization
• physics-guided reward
• semantic reward
• surgical smoke removal