TRCoRSurg: Temporal-Relational Co-Reasoning for Surgical Video Triplet Recognition

2026-08-05

AI Summary

This work addresses the challenge that existing methods for surgical video triplet recognition struggle to jointly model intra-frame label dependencies and inter-frame temporal semantics, often yielding inconsistent relationships. To overcome this, the authors propose a unified framework that integrates spatial, relational, and temporal cues: a multi-scale encoder captures class-specific spatial priors, while a novel Multi-Scale Class Activation Map–guided Relation Extraction module (MS-CAMRE) and a Bidirectional Temporal–Relation Fusion Attention mechanism (BTRFA) are introduced to enhance relation modeling and temporal coherence. Additionally, a new evaluation metric, Temporal Causal and Semantic Consistency Error Rate (TCER), is proposed to quantify causal and semantic consistency in predicted triplets. Experiments on CholecT45 and ProstaTD datasets demonstrate state-of-the-art performance, with AP_IVT improvements of 5.1% and 7.8%, respectively, and TCER reductions exceeding 36% and 25%.

Abstract

Understanding complex surgical scenes requires recognizing multiple interdependent entities, such as instruments, actions, and targets, while maintaining their relational consistency across time. Existing surgical triplet recognition methods struggle to jointly model intra-frame label dependencies and inter-frame temporal semantics in a unified manner. To address these limitations, we propose a unified framework that integrates spatial, relational, and temporal cues for robust surgical triplet recognition. Specifically, class-specific spatial priors are first extracted through a multi-scale encoder. These priors are then refined by a Label Correlation Modeling module with multi-scale class activation map-guided relational extraction (MS-CAMRE), enabling the model to capture both static co-occurrence patterns and dynamic contextual dependencies among triplet components. Furthermore, a Bidirectional Temporal-Relational Fusion Attention (BTRFA) module harmonizes temporal and relational representations to achieve coherent temporal reasoning. We also introduce a new evaluation metric, the Triplet Consistency Error Rate (TCER), which quantitatively measures the model's ability to preserve causal and semantic consistency across triplets. Extensive experiments on the CholecT45 and ProstaTD datasets show that our method achieves state-of-the-art performance, improving AP_IVT by 5.1 percent and 7.8 percent, respectively. Moreover, according to TCER, our approach achieves relative reductions of more than 36 percent and 25 percent on the two datasets, respectively, demonstrating the effectiveness of our framework in temporal-relational co-reasoning.

Problem

• surgical triplet recognition
• temporal semantics
• relational consistency
• label dependencies
• surgical video understanding

Innovation

• Temporal-Relational Co-Reasoning
• Surgical Triplet Recognition
• Multi-Scale CAM-Guided Relational Extraction
• Bidirectional Temporal-Relational Fusion
• Triplet Consistency Error Rate