Towards Polyp Counting In Full-Procedure Colonoscopy Videos

2025-02-14

AI Summary

This work addresses the challenge of automated polyp counting in full-colonoscopy videos, targeting three key issues: polyp trajectory fragmentation, cross-frame re-identification (ReID) difficulty, and duplicate counting—critical for computing quality metrics such as adenoma detection rate (ADR) and polyp per colonoscopy (PPC). Methodologically, we establish the first end-to-end polyp counting benchmark on the REAL-Colon open dataset; innovatively integrate SimCLR-based self-supervised learning to jointly model single-frame and multi-view trajectory representations; and enhance trajectory re-association via trajectory-level ReID coupled with Affinity Propagation clustering. Our approach significantly mitigates fragmentation while suppressing false positives. Evaluated on REAL-Colon, it achieves a polyp fragmentation rate of 6.30 and a false positive rate <5%, setting a new state-of-the-art. The framework provides a reproducible, robust foundation for automated colonoscopy quality control.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multimodal Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

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
• Security and Privacy: Data transparency and provenance

Abstract

Automated colonoscopy reporting holds great potential for enhancing quality control and improving cost-effectiveness of colonoscopy procedures. A major challenge lies in the automated identification, tracking, and re-association (ReID) of polyps tracklets across full-procedure colonoscopy videos. This is essential for precise polyp counting and enables automated computation of key quality metrics, such as Adenoma Detection Rate (ADR) and Polyps Per Colonoscopy (PPC). However, polyp ReID is challenging due to variations in polyp appearance, frequent disappearance from the field of view, and occlusions. In this work, we leverage the REAL-Colon dataset, the first open-access dataset providing full-procedure videos, to define tasks, data splits and metrics for the problem of automatically count polyps in full-procedure videos, establishing an open-access framework. We re-implement previously proposed SimCLR-based methods for learning representations of polyp tracklets, both single-frame and multi-view, and adapt them to the polyp counting task. We then propose an Affinity Propagation-based clustering method to further improve ReID based on these learned representations, ultimately enhancing polyp counting. Our approach achieves state-of-the-art performance, with a polyp fragmentation rate of 6.30 and a false positive rate (FPR) below 5% on the REAL-Colon dataset. We release code at https://github.com/lparolari/towards-polyp-counting.

Problem

• Automated polyp identification and tracking
• Enhancing polyp counting accuracy
• Improving colonoscopy quality metrics computation

Innovation

• SimCLR-based representation learning
• Affinity Propagation clustering
• Real-Colon dataset utilization