A Neighborhood Attention Transformer Network for Enhanced 3D Segmentation of the Left Anterior Descending Artery

2026-08-12

AI Summary

This study addresses the challenging segmentation of the left anterior descending (LAD) artery in non-contrast, free-breathing 3D CT scans, where its small size, low soft-tissue contrast, and high anatomical variability hinder accurate delineation. To tackle this, the authors propose NA-UNETR, a 3D Transformer-based architecture that integrates neighborhood attention and dilated neighborhood attention modules to jointly capture local details and global context. The model employs LoRA for parameter-efficient transfer learning and introduces a novel homoscedastic uncertainty-weighted composite loss combining Dice-Focal and Hausdorff distance to dynamically refine boundary accuracy. Evaluated under extremely limited annotation, NA-UNETR achieves a Dice score of 45.64% on in-house data and significantly improves to 79.49% on the ImageCAS dataset, outperforming nnU-Net and Swin UNETR in boundary metrics (HD95 = 38.16 mm, ASD = 10.01 mm).

Abstract

Background: Accurate segmentation of the Left Anterior Descending (LAD) artery in 3D free-breathing, non-contrast CT is critical for cardiac dose sparing in thoracic radiotherapy. The LAD is extremely small, has poor soft-tissue contrast, and varies substantially across patients; even manual contours show limited inter-observer agreement, underscoring the ambiguity of the vessel boundaries. Purpose: To develop a transformer-based framework that improves LAD delineation in low-contrast, imbalanced CT through local-global context modeling and uncertainty-guided optimization. Methods: We propose NA-UNETR, a 3D transformer-based segmentation model whose Neighborhood Attention (NA) and Dilated NA (DiNA) blocks jointly capture fine structural detail and long-range context. Given the scarcity of annotated LAD data, the model is pretrained on 1,000 CTA volumes of general coronary anatomy and fine-tuned with LoRA-based parameter-efficient adaptation on 20 free-breathing institutional CT scans. A composite Dice-Focal and Hausdorff loss, dynamically balanced via homoscedastic uncertainty, improves overlap and boundary accuracy. Results: NA-UNETR reached 45.64% Dice, 38.16 mm HD95, and 10.01 mm ASD, improving Dice by 3.10 percentage points over nnU-Net and reducing HD95 by 2.96 mm relative to Swin UNETR, with the strongest boundary accuracy among all models and improved centerline stability. On ImageCAS it achieved 79.49% Dice, 8.89 mm HD95, and 1.02 mm ASD. Ablations confirmed that residual blocks, variable kernels, and uncertainty-weighted loss each contributed. Conclusions: NA-UNETR balances local precision and global context for thin, low-contrast LAD structures, offering a computationally efficient framework for substructure-level cardiac segmentation in radiotherapy planning.

Problem

• Left Anterior Descending artery
• 3D segmentation
• low-contrast CT
• boundary ambiguity
• cardiac radiotherapy

Innovation

• Neighborhood Attention
• Transformer-based segmentation
• Uncertainty-guided loss
• Parameter-efficient fine-tuning
• Low-contrast 3D segmentation