Adaptive Frequency Domain Alignment Network for Medical image segmentation

2025-12-18

AI Summary

To address the scarcity of labeled data in the target domain for medical image segmentation, this paper proposes an unsupervised domain adaptation framework based on frequency-domain adaptive feature alignment. The method introduces a novel Fourier-domain adversarial alignment mechanism that jointly models spatial and frequency representations, overcoming the limitations of conventional spatial-domain alignment. It comprises three core components: (1) Fourier-domain adversarial training, (2) multi-scale spatial-frequency feature interaction, and (3) an unsupervised domain adaptation strategy. Evaluated on the self-constructed VITILIGO2025 dataset, the framework achieves 90.9% IoU; on the DRIVE retinal vessel segmentation benchmark, it attains 82.6% IoU—both significantly surpassing current state-of-the-art methods. These results empirically validate the efficacy and robustness of frequency-domain modeling for domain transfer in medical imaging.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Application Domains: Mobility, Driving & Flight

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Security and Privacy: Data transparency and provenance
• User Modeling, Personalization and Recommendation: User modeling for targeted and personalized online advertising

Abstract

High-quality annotated data plays a crucial role in achieving accurate segmentation. However, such data for medical image segmentation are often scarce due to the time-consuming and labor-intensive nature of manual annotation. To address this challenge, we propose the Adaptive Frequency Domain Alignment Network (AFDAN)--a novel domain adaptation framework designed to align features in the frequency domain and alleviate data scarcity. AFDAN integrates three core components to enable robust cross-domain knowledge transfer: an Adversarial Domain Learning Module that transfers features from the source to the target domain; a Source-Target Frequency Fusion Module that blends frequency representations across domains; and a Spatial-Frequency Integration Module that combines both frequency and spatial features to further enhance segmentation accuracy across domains. Extensive experiments demonstrate the effectiveness of AFDAN: it achieves an Intersection over Union (IoU) of 90.9% for vitiligo segmentation in the newly constructed VITILIGO2025 dataset and a competitive IoU of 82.6% on the retinal vessel segmentation benchmark DRIVE, surpassing existing state-of-the-art approaches.

Problem

• Addresses medical image segmentation data scarcity
• Aligns features in frequency domain for adaptation
• Enhances cross-domain segmentation accuracy via fusion

Innovation

• Adaptive frequency domain alignment for feature matching
• Adversarial learning module transfers source to target domain
• Spatial-frequency integration enhances cross-domain segmentation accuracy