From Lines to Shapes: Geometric-Constrained Segmentation of X-Ray Collimators via Hough Transform

2025-09-04

AI Summary

In X-ray imaging, scatter radiation blurs collimator shadow edges, severely hindering accurate segmentation of polygonal regions of interest (ROIs). To address this, we propose a novel geometrically informed deep learning framework. Our method introduces a differentiable Hough transform module that jointly optimizes edge detection and ROI centroid localization; incorporates a line-constrained segmentation mechanism that explicitly encodes polygonal geometry in parameter space; and employs multi-task learning to enhance robustness. Crucially, the approach requires no prior specification of vertex count, enabling generalization to arbitrary polygonal collimation fields. Evaluated on real clinical X-ray images, it achieves a median Hausdorff distance of 4.3–5.0 mm—substantially outperforming conventional methods—while demonstrating high accuracy, strong generalizability across diverse collimation geometries, and direct clinical applicability.

Technology Category

• Computer Vision: Segmentation
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Security and Privacy: Data transparency and provenance
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Collimation in X-ray imaging restricts exposure to the region-of-interest (ROI) and minimizes the radiation dose applied to the patient. The detection of collimator shadows is an essential image-based preprocessing step in digital radiography posing a challenge when edges get obscured by scattered X-ray radiation. Regardless, the prior knowledge that collimation forms polygonal-shaped shadows is evident. For this reason, we introduce a deep learning-based segmentation that is inherently constrained to its geometry. We achieve this by incorporating a differentiable Hough transform-based network to detect the collimation borders and enhance its capability to extract the information about the ROI center. During inference, we combine the information of both tasks to enable the generation of refined, line-constrained segmentation masks. We demonstrate robust reconstruction of collimated regions achieving median Hausdorff distances of 4.3-5.0mm on diverse test sets of real Xray images. While this application involves at most four shadow borders, our method is not fundamentally limited by a specific number of edges.

Problem

• Detect obscured collimator shadows in X-ray images
• Segment polygonal-shaped shadows with geometric constraints
• Reconstruct collimated regions to minimize radiation exposure

Innovation

• Deep learning segmentation with geometric constraints
• Differentiable Hough transform network for border detection
• Combined task information for refined mask generation