arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2602.19322cs.CVcs.AIcs.LG
收藏

US-JEPA:一种用于医学超声的联合嵌入预测架构

US-JEPA: A Joint Embedding Predictive Architecture for Ultrasound

Biomedical AI Research Lab, University of California, Los Angeles, USA(生物医学人工智能研究实验室,加州大学洛杉矶分校);Department of Bioengineering, University of California, Los Angeles, USA(生物工程系,加州大学洛杉矶分校);Medical Informatics Home Area, University of California, Los Angeles, USA(医学信息学家庭区域,加州大学洛杉矶分校);Department of Electrical and Computer Engineering, University of California, Los Angeles, USA(电气与计算机工程系,加州大学洛杉矶分校);Department of Pathology, University of California, Los Angeles, USA(病理学系,加州大学洛杉矶分校);Department of Radiology, University of California, Los Angeles, USA(放射学系,加州大学洛杉矶分校)

Ashwath Radhachandran, Vedrana Ivezić, Shreeram Athreya, Corey W. Arnold, William Speier

AI总结:

US-JEPA提出了一种自监督框架,通过静态教师不对称潜在训练目标,提升超声图像的稳健表示能力。

摘要:

超声成像在表示学习中面临独特挑战,因其固有的噪声采集过程。低信噪比和随机斑点图案阻碍了依赖像素级重建目标的标准自监督学习方法。联合嵌入预测架构(JEPAs)通过预测掩码的潜在表示而不是原始像素来解决这一缺点。然而,标准方法依赖于超参数敏感且计算成本高的在线教师,通过指数移动平均更新。我们提出了US-JEPA,一种自监督框架,采用静态教师不对称潜在训练(SALT)目标。通过使用冻结的领域特定教师来提供稳定的潜在目标,US-JEPA将学生-教师优化解耦,并促使学生扩展教师的语义先验。此外,我们提供了所有公开可用的最新超声基础模型在UltraBench公共数据集基准上的首次严谨比较。在针对多样化分类任务的线性探测中,US-JEPA在性能上与或优于领域特定和通用视觉基础模型基线。我们的结果表明,掩码潜在预测提供了一条稳定且高效的通往稳健超声表示的道路。

英文摘要:

Ultrasound (US) imaging poses unique challenges for representation learning due to its inherently noisy acquisition process. The low signal-to-noise ratio and stochastic speckle patterns hinder standard self-supervised learning methods relying on a pixel-level reconstruction objective. Joint-Embedding Predictive Architectures (JEPAs) address this drawback by predicting masked latent representations rather than raw pixels. However, standard approaches depend on hyperparameter-brittle and computationally expensive online teachers updated via exponential moving average. We propose US-JEPA, a self-supervised framework that adopts the Static-teacher Asymmetric Latent Training (SALT) objective. By using a frozen, domain-specific teacher to provide stable latent targets, US-JEPA decouples student-teacher optimization and pushes the student to expand upon the semantic priors of the teacher. In addition, we provide the first rigorous comparison of all publicly available state-of-the-art ultrasound foundation models on UltraBench, a public dataset benchmark spanning multiple organs and pathological conditions. Under linear probing for diverse classification tasks, US-JEPA achieves performance competitive with or superior to domain-specific and universal vision foundation model baselines. Our results demonstrate that masked latent prediction provides a stable and efficient path toward robust ultrasound representations.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑