DAUNet: A Lightweight UNet Variant with Deformable Convolutions and Parameter-Free Attention for Medical Image Segmentation

2025-12-07

AI Summary

To address the critical need for lightweight, spatially adaptive, and context-aware models in resource-constrained clinical medical image segmentation, this paper proposes an efficient UNet variant. Methodologically, it integrates deformable convolutional networks v2 (DCNv2) to dynamically model geometric deformations, incorporates the parameter-free SimAM attention mechanism for saliency-aware feature enhancement, and introduces a lightweight bottleneck module to optimize encoder-decoder feature fusion. The resulting architecture achieves high inference efficiency and low parameter count while significantly improving robustness to low-contrast regions and anatomical deformations. Quantitative evaluation on the FH-PS-AoP and FUMPE datasets demonstrates superior performance over state-of-the-art methods across key metrics—Dice score, 95th-percentile Hausdorff distance (HD95), and average surface distance (ASD)—validating its high parameter efficiency, strong generalizability, and clinical applicability.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Calibration & Uncertainty Quantification
• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Medical image segmentation plays a pivotal role in automated diagnostic and treatment planning systems. In this work, we present DAUNet, a novel lightweight UNet variant that integrates Deformable V2 Convolutions and Parameter-Free Attention (SimAM) to improve spatial adaptability and context-aware feature fusion without increasing model complexity. DAUNet's bottleneck employs dynamic deformable kernels to handle geometric variations, while the decoder and skip pathways are enhanced using SimAM attention modules for saliency-aware refinement. Extensive evaluations on two challenging datasets, FH-PS-AoP (fetal head and pubic symphysis ultrasound) and FUMPE (CT-based pulmonary embolism detection), demonstrate that DAUNet outperforms state-of-the-art models in Dice score, HD95, and ASD, while maintaining superior parameter efficiency. Ablation studies highlight the individual contributions of deformable convolutions and SimAM attention. DAUNet's robustness to missing context and low-contrast regions establishes its suitability for deployment in real-time and resource-constrained clinical environments.

Problem

• Improves medical image segmentation with lightweight UNet variant
• Enhances spatial adaptability using deformable convolutions and attention
• Achieves robust performance in real-time clinical environments

Innovation

• Deformable convolutions handle geometric variations adaptively
• Parameter-free SimAM attention enables saliency-aware feature refinement
• Lightweight UNet variant maintains efficiency while improving segmentation accuracy