发表机构
National Yang Ming Chiao Tung University; Ho Chi Minh City University of Technology and Engineering（国立阳明交通大学; 胡志明市理工大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对手持超声图像复合退化问题，提出两阶段框架：先通过循环一致风格迁移生成像素对齐数据，再用DDG-LoRA微调增强模型，在USenhance2023上FID提升16.7%。
AI 中文摘要
低成本手持超声设备相比专业医院超声机器可以广泛部署。然而，其图像遭受复合退化，可能误导临床判断。受此观察启发，将手持低质量（LQ）图像映射到医院高质量（HQ）图像已被认为是一个有价值的研究问题。传统上，这种映射需要像素对齐的LQ-HQ对。该要求在实际场景中不令人满意，因为不同时间的真实扫描永远不会像素对齐。本文通过一个两阶段框架解决这一挑战。第一阶段生成像素对齐的LQ-HQ数据集，第二阶段训练一个增强模型以改善LQ图像。第一阶段在未对齐的真实LQ-HQ对上训练一个循环一致风格迁移模型，以学习HQ-to-LQ模型。然后，该模型将真实HQ图像转换为像素对齐的LQ图像。基于第一阶段生成的数据集，第二阶段使用双退化引导（DDG）低秩适配（LoRA）方法，基于对齐对微调一个LQ-to-HQ模型。在此阶段，模型基于众所周知的PiSA-SR框架，但插入了一个退化条件校正矩阵。在USenhance2023数据集上的实验结果表明，FID指标相比最强基线提高了16.7%，而其他指标表明我们的增强输出与真实HQ分布良好对齐。我们方法的源代码可在该https URL获取。
英文摘要
Low-cost handheld ultrasound devices can be widely deployed compared to professional hospital ultrasound machines. However, their images suffer from compound degradation that can mislead clinical judgment. Motivated by this observation, mapping handheld low-quality (LQ) to hospital high-quality (HQ) images has been considered a valuable research question. Conventionally, the mapping requires pixel-aligned LQ-HQ pairs. This requirement is unsatisfactory in practical scenarios because real scans at different times are never pixel-aligned. This paper addresses the challenge with a two-stage framework. The first stage generates pixel-aligned LQ-HQ datasets, and the second stage trains an enhancement model that improves LQ images. The first stage trains a cycle-consistent style-transfer model on unaligned real LQ-HQ pairs to learn a HQ-to-LQ model. Then, the model transforms real HQ images into pixel-aligned LQ images. Based on the dataset generated by the first stage, the second stage uses the Dual Degradation-Guided (DDG) Low-Rank Adaptation (LoRA) method to fine-tune an LQ-to-HQ model based on aligned pairs. In this stage, the model is based on the well known PiSA-SR framework but inserts a degradation-conditioned correction matrix. Experimental results on the USenhance2023 dataset show that the FID metric is improved by 16.7% over the strongest baseline while other metrics indicate that our enhanced outputs are well aligned with the real HQ distribution. The source code of our method is available at https://github.com/Jason0411202/DDG_LoRA.
