CPKD: Clinical Prior Knowledge-Constrained Diffusion Models for Surgical Phase Recognition in Endoscopic Submucosal Dissection

TL;DR

This paper introduces CPKD, a diffusion model-based framework that leverages clinical prior knowledge for improved surgical phase recognition in endoscopic procedures, addressing limitations of existing iterative refinement methods.

Contribution

The paper presents a novel diffusion-based generative approach incorporating clinical prior knowledge and domain-specific strategies for more accurate surgical phase recognition.

Findings

CPKD outperforms existing methods on multiple datasets.

Incorporating clinical prior knowledge enhances model accuracy.

Diffusion models are effective for temporal sequence recognition in surgery.

Abstract

Gastrointestinal malignancies constitute a leading cause of cancer-related mortality worldwide, with advanced-stage prognosis remaining particularly dismal. Originating as a groundbreaking technique for early gastric cancer treatment, Endoscopic Submucosal Dissection has evolved into a versatile intervention for diverse gastrointestinal lesions. While computer-assisted systems significantly enhance procedural precision and safety in ESD, their clinical adoption faces a critical bottleneck: reliable surgical phase recognition within complex endoscopic workflows. Current state-of-the-art approaches predominantly rely on multi-stage refinement architectures that iteratively optimize temporal predictions. In this paper, we present Clinical Prior Knowledge-Constrained Diffusion (CPKD), a novel generative framework that reimagines phase recognition through denoising diffusion principles while preserving the core iterative refinement philosophy. This architecture progressively reconstructs phase sequences starting from random noise and conditioned on visual-temporal features. To better capture three domain-specific characteristics, including positional priors, boundary ambiguity, and relation dependency, we design a conditional masking strategy. Furthermore, we incorporate clinical prior knowledge into the model training to improve its ability to correct phase logical errors. Comprehensive evaluations on ESD820, Cholec80, and external multi-center demonstrate that our proposed CPKD achieves superior or comparable performance to state-of-the-art approaches, validating the effectiveness of diffusion-based generative paradigms for surgical phase recognition.