Boundary-Emphasized Weight Maps for Distal Airway Segmentation

2025-02-28

AI Summary

This work addresses three critical challenges in automatic distal airway segmentation from pulmonary CT: boundary leakage, branch discontinuity, and severe class imbalance. To this end, we propose Boundary Enhancement Loss (BEL), a novel voxel-wise boundary detection–driven loss function that replaces conventional centerline-guided paradigms. BEL incorporates an adaptive weighting mechanism to jointly optimize boundary localization accuracy and topological structure preservation. Integrated into a 3D U-Net architecture, BEL embeds boundary priors into a weighted cross-entropy formulation. Evaluated on the ATM22 and AIIB23 benchmarks, our method achieves substantial improvements in topological metrics—e.g., branch recall (BR) and connectivity (RC)—and boosts small-airway segmentation accuracy by 12.7%. Both under-segmentation and over-segmentation errors are significantly reduced, effectively overcoming the long-standing bottleneck in resolving fine distal branches.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Calibration & Uncertainty Quantification
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Automated airway segmentation from lung CT scans is vital for diagnosing and monitoring pulmonary diseases. Despite advancements, challenges like leakage, breakage, and class imbalance persist, particularly in capturing small airways and preserving topology. We propose the Boundary-Emphasized Loss (BEL), which enhances boundary preservation using a boundary-based weight map and an adaptive weight refinement strategy. Unlike centerline-based approaches, BEL prioritizes boundary voxels to reduce misclassification, improve topology, and enhance structural consistency, especially on distal airway branches. Evaluated on ATM22 and AIIB23, BEL outperforms baseline loss functions, achieving higher topology-related metrics and comparable overall-based measures. Qualitative results further highlight BEL's ability to capture fine anatomical details and reduce segmentation errors, particularly in small airways. These findings establish BEL as a promising solution for accurate and topology-enhancing airway segmentation in medical imaging.

Problem

• Improves segmentation of small airways in lung CT scans
• Reduces leakage and breakage in airway segmentation
• Enhances topology preservation and structural consistency

Innovation

• Boundary-Emphasized Loss enhances airway segmentation accuracy
• Adaptive weight refinement strategy improves topology preservation
• Prioritizes boundary voxels to reduce misclassification errors