TanGo：面向医学图像泛化的分布级基础先验一致性与实例级风格校准学习
TANGO: Learning Distribution-wise Foundation Prior Consistency and Instance-wise Style Calibration for Medical Image Generalization
把冻结的视觉基础模型编码器中的低频稳定表征作为分布级先验来约束源模型的特征分布，并用可学习的数据装饰器把测试图像拉回增强后的源域分布——这套「基础模型先验 + 实例级风格校准」的组合可直接迁移到其他面临域偏移的密集预测任务（如跨中心分割、跨域检测）。
原文依据 We incorporate stable low-frequency representations from a frozen encoder of VFMs as priors to guide the source model
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论测试时自适应已成为应对医学图像分割中真实世界域偏移的一种有前景的方案。现有方法通过更新或正则化一个预训练的源模型来实现自适应。然而，它们面临两个主要问题：(i) 其所依赖的源模型在域偏移下容易过拟合；(ii) 在动态持续测试场景中，误差累积与类别遗忘会被进一步加剧。为克服这些局限，我们提出 TanGo，一个将「为自适应而训练」与基础引导及持续风格校准相结合的新框架。在训练阶段，TanGo 通过分布级一致性学习从视觉基础模型中学习泛化先验。我们将视觉基础模型冻结编码器中的稳定低频表征作为先验来引导源模型，约束其输出特征分布，从而获得更具泛化性的特征空间。在测试阶段，我们提出一种实例级风格校准方法，使用可学习的数据装饰器将动态测试图像拉回增强后的类源分布。随后施加一系列约束，使装饰后的测试样本更接近增强后的源分布，同时保持其语义完整性，从而提升源模型的持续适应能力。在多个医学图像分割任务上的大量实验表明，TanGo 取得了最优性能。
英文原文摘要 1492 字符
Test-time adaptation (TTA) has emerged as a promising solution to address real world domain shifts in medical image segmentation. Current approaches adapt by updating or regularizing a pre-trained source model. However, they face two major issues: (i) the source models on which they rely are prone to overfitting under domain shifts; (ii) in dynamic continual testing scenarios, error accumulation and class forgetting are further exacerbated. To overcome these limitations, we propose TanGo, a novel framework that combines Training to adapt with Foundation Guidance and continual style calibration. During training, TanGo learns generalization priors from vision foundation models (VFMs) through distribution-wise consistency learning. We incorporate stable low-frequency representations from a frozen encoder of VFMs as priors to guide the source model, constraining its output feature distribution to yield a more generalizable feature space. At test time, we propose an instance-wise style calibration method that uses a learnable data decorator to shift dynamic test images back toward an enhanced source-like distribution. Subsequently, a series of constraints is applied to draw the decorated test samples closer to the enhanced source distribution while maintaining their semantic integrity, thus boosting the continual adaptability of the source model. Extensive experiments on multiple medical image segmentation tasks demonstrate that TanGo achieves state-of-the-art performance.