UI-Styler: Ultrasound Image Style Transfer with Class-Aware Prompts for Cross-Device Diagnosis Using a Frozen Black-Box Inference Network

2025-11-21

AI Summary

Ultrasound images exhibit significant domain shift across different acquisition devices, severely limiting the cross-device generalization of frozen black-box downstream models (e.g., classification and segmentation networks). Existing unpaired image translation methods lack semantic class constraints, leading to content–class mismatches that compromise diagnostic reliability. To address this, we propose a class-aware prompt-guided style transfer framework: it generates class-aware prompts via pseudo-labeling and aligns device-specific texture distributions between domains through a texture pattern matching mechanism—while preserving source-domain anatomical structure. The method operates without paired data and is fully compatible with arbitrary frozen black-box inference models. Evaluated on multiple cross-device ultrasound tasks, it substantially reduces inter-domain distribution divergence, achieving state-of-the-art accuracy in both classification and segmentation. This significantly enhances robustness for clinical deployment.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Computer Vision: Medical and Biological Imaging
• Humans and AI: Interaction Techniques and Devices

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

The appearance of ultrasound images varies across acquisition devices, causing domain shifts that degrade the performance of fixed black-box downstream inference models when reused. To mitigate this issue, it is practical to develop unpaired image translation (UIT) methods that effectively align the statistical distributions between source and target domains, particularly under the constraint of a reused inference-blackbox setting. However, existing UIT approaches often overlook class-specific semantic alignment during domain adaptation, resulting in misaligned content-class mappings that can impair diagnostic accuracy. To address this limitation, we propose UI-Styler, a novel ultrasound-specific, class-aware image style transfer framework. UI-Styler leverages a pattern-matching mechanism to transfer texture patterns embedded in the target images onto source images while preserving the source structural content. In addition, we introduce a class-aware prompting strategy guided by pseudo labels of the target domain, which enforces accurate semantic alignment with diagnostic categories. Extensive experiments on ultrasound cross-device tasks demonstrate that UI-Styler consistently outperforms existing UIT methods, achieving state-of-the-art performance in distribution distance and downstream tasks, such as classification and segmentation.

Problem

• Ultrasound images vary across devices causing domain shifts
• Existing methods lack class-aware alignment during style transfer
• Need to preserve diagnostic accuracy while adapting image styles

Innovation

• Class-aware prompts guide ultrasound image style transfer
• Pattern-matching transfers target textures while preserving source structure
• Frozen black-box network enables cross-device diagnosis without retraining