Uncertainty-Gated Deformable Network for Breast Tumor Segmentation in MR Images

2025-09-19

AI Summary

To address the challenges of modeling irregular tumor morphology and insufficient integration of local and global features in breast MRI segmentation, this paper proposes the Uncertainty-Gated Deformable Network (UGDN). UGDN innovatively combines the local receptive capability of CNNs with the long-range dependency modeling strength of Transformers. It introduces an uncertainty-gated module for pixel-wise selective feature interaction and integrates deformable convolution with attention mechanisms to adaptively capture complex tumor boundaries. Furthermore, a boundary-sensitive deep supervision loss is designed to enhance contour delineation accuracy. Evaluated on two clinical breast MRI datasets, UGDN consistently outperforms state-of-the-art methods—including TransUNet and Swin-Unet—achieving Dice score improvements of 2.3–3.1%. These results demonstrate UGDN’s effectiveness and generalizability for precise, clinically applicable tumor segmentation.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Calibration & Uncertainty Quantification
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization

Abstract

Accurate segmentation of breast tumors in magnetic resonance images (MRI) is essential for breast cancer diagnosis, yet existing methods face challenges in capturing irregular tumor shapes and effectively integrating local and global features. To address these limitations, we propose an uncertainty-gated deformable network to leverage the complementary information from CNN and Transformers. Specifically, we incorporates deformable feature modeling into both convolution and attention modules, enabling adaptive receptive fields for irregular tumor contours. We also design an Uncertainty-Gated Enhancing Module (U-GEM) to selectively exchange complementary features between CNN and Transformer based on pixel-wise uncertainty, enhancing both local and global representations. Additionally, a Boundary-sensitive Deep Supervision Loss is introduced to further improve tumor boundary delineation. Comprehensive experiments on two clinical breast MRI datasets demonstrate that our method achieves superior segmentation performance compared with state-of-the-art methods, highlighting its clinical potential for accurate breast tumor delineation.

Problem

• Segments irregular breast tumor shapes in MRI
• Integrates local and global features effectively
• Improves tumor boundary delineation accuracy

Innovation

• Deformable feature modeling for adaptive receptive fields
• Uncertainty-gated module for selective feature exchange
• Boundary-sensitive loss for improved tumor delineation