Dual Interaction Network with Cross-Image Attention for Medical Image Segmentation

2025-09-07

AI Summary

Medical image segmentation is often degraded by noise, blur, and low contrast, leading to inaccurate lesion boundary localization. Conventional enhancement methods tend to introduce artifacts or compromise diagnostically critical information, while traditional fusion strategies (e.g., feature concatenation) struggle to simultaneously achieve complementarity and robustness. To address these challenges, we propose a Dual Interactive Fusion Module that establishes a bidirectional cross-image attention mechanism, enabling dynamic complementary modeling between original and enhanced images across multiple feature levels; additionally, a global spatial attention module refines feature representation. Our network adopts a dual-branch architecture, integrating cross-attention and a multi-scale boundary gradient loss for end-to-end joint segmentation. Evaluated on the ACDC and Synapse datasets, our method achieves significant improvements in segmentation accuracy—particularly in boundary preservation—outperforming state-of-the-art approaches both quantitatively and qualitatively.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Multi-instance/Multi-view Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization

Abstract

Medical image segmentation is a crucial method for assisting professionals in diagnosing various diseases through medical imaging. However, various factors such as noise, blurriness, and low contrast often hinder the accurate diagnosis of diseases. While numerous image enhancement techniques can mitigate these issues, they may also alter crucial information needed for accurate diagnosis in the original image. Conventional image fusion strategies, such as feature concatenation can address this challenge. However, they struggle to fully leverage the advantages of both original and enhanced images while suppressing the side effects of the enhancements. To overcome the problem, we propose a dual interactive fusion module (DIFM) that effectively exploits mutual complementary information from the original and enhanced images. DIFM employs cross-attention bidirectionally to simultaneously attend to corresponding spatial information across different images, subsequently refining the complementary features via global spatial attention. This interaction leverages low- to high-level features implicitly associated with diverse structural attributes like edges, blobs, and object shapes, resulting in enhanced features that embody important spatial characteristics. In addition, we introduce a multi-scale boundary loss based on gradient extraction to improve segmentation accuracy at object boundaries. Experimental results on the ACDC and Synapse datasets demonstrate the superiority of the proposed method quantitatively and qualitatively. Code available at: https://github.com/JJeong-Gari/DIN

Problem

• Improving medical image segmentation accuracy despite noise and low contrast
• Leveraging complementary information from original and enhanced medical images
• Enhancing boundary segmentation through multi-scale gradient-based loss

Innovation

• Dual interactive fusion module with cross-attention
• Bidirectional cross-image spatial attention mechanism
• Multi-scale boundary loss using gradient extraction