[论文] Medical Image Segmentation: Foundation Model Adaptation versus Task-Sp...

论文概要

研究领域: CV
作者: Emre Kutlusan, Yongmin Li
发布时间: 2026-11-10
arXiv: 2611.14519

中文摘要

医学图像分割领域，越来越多人将自然图像上预训练的基础模型适配到医学任务，期望取代按任务训练的模型。然而现有比较大多使用训练分布内的数据，且仅与一个配置未必合理的有监督基线对比，对新临床站点的证据有限。也不清楚基础模型究竟需要适配多少、由什么决定。我们将 DINOv2 以从仅调节偏置项到全量微调的多种策略适配，与从零训练的 U-Net 和 ResU-Net 对比，覆盖结肠镜、乳腺超声和皮肤镜，并在训练源留出数据与其他机构的独立数据集上测试。适配后的基础模型在所有外部数据集上泛化更好；训练分布内，数据充足时任务专用模型仍有竞争力且边界刻画更精细，基础模型更易漏检小病灶。基础模型只需适配很小一部分容量，所需容量由成像域而非适配方法决定。我们还发现一个广泛使用的息肉测试集与另一基准存在图像重复。结论：跨站点场景下适配基础模型是值得的，容量应基于目标域选择。

原文摘要

Foundation models pretrained on large natural-image collections are increasingly adapted for medical image segmentation, in the expectation that they will replace models trained for each task. Most comparisons, however, use data from the training distribution and a single task-specific baseline whose configuration is rarely justified, and therefore provide limited evidence of performance at new clinical sites. It also remains unclear how much of a foundation model must be adapted, and what determines this. We adapted a foundation model, DINOv2, with strategies ranging from tuning only its bias terms to full fine-tuning, and compared it with U-Nets and ResU-Nets trained from scratch. Experiments covered colonoscopy, breast ultrasound and dermoscopy, with testing on held-out data from the tr...

自动采集于 2026-11-12

#论文 #arXiv #CV #小凯