RepAir: A Framework for Airway Segmentation and Discontinuity Correction in CT

2025-11-18

AI Summary

To address the prevalent topological disconnection issue in automatic airway segmentation from CT images, this paper proposes a three-stage anatomy-aware segmentation framework. First, an initial segmentation is generated using nnU-Net. Second, candidate connection paths are extracted based on the centerline skeleton. Third, a lightweight 1D convolutional classifier discriminates true branches from spurious connections, enabling skeleton-guided topological repair. The method preserves high voxel-wise accuracy while significantly improving airway tree connectivity and anatomical plausibility. Evaluated on the ATM’22 and AeroPath datasets, it outperforms state-of-the-art methods—including Bronchinet and NaviAirway—across key metrics: Dice score, average surface distance (ASD), branch connectivity rate, and topological integrity. This yields more reliable and interpretable airway representations for quantitative pulmonary analysis.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Calibration & Uncertainty Quantification
• Knowledge Representation and Reasoning: Automated Reasoning and Theorem Proving

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Web Mining and Content Analysis: Web data integration and cleaning

Abstract

Accurate airway segmentation from chest computed tomography (CT) scans is essential for quantitative lung analysis, yet manual annotation is impractical and many automated U-Net-based methods yield disconnected components that hinder reliable biomarker extraction. We present RepAir, a three-stage framework for robust 3D airway segmentation that combines an nnU-Net-based network with anatomically informed topology correction. The segmentation network produces an initial airway mask, after which a skeleton-based algorithm identifies potential discontinuities and proposes reconnections. A 1D convolutional classifier then determines which candidate links correspond to true anatomical branches versus false or obstructed paths. We evaluate RepAir on two distinct datasets: ATM'22, comprising annotated CT scans from predominantly healthy subjects and AeroPath, encompassing annotated scans with severe airway pathology. Across both datasets, RepAir outperforms existing 3D U-Net-based approaches such as Bronchinet and NaviAirway on both voxel-level and topological metrics, and produces more complete and anatomically consistent airway trees while maintaining high segmentation accuracy.

Problem

• Automated airway segmentation produces disconnected components in CT scans
• Discontinuities hinder reliable biomarker extraction for lung analysis
• Existing methods fail to maintain anatomical consistency across pathologies

Innovation

• U-Net-based network for initial airway segmentation
• Skeleton-based algorithm identifies potential discontinuities
• 1D convolutional classifier determines true anatomical branches