DINOv2 适配 vs 专用训练：医学影像分割跨机构泛化对比研究
Medical Image Segmentation: Foundation Model Adaptation versus Task-Specific Training
做医学影像的人该看看这篇：他们拿 DINOv2 和 U-Net 在三种影像上正面对比，还测了跨医院的外部数据，结论很实在——院内任务用老方法就行，要跨机构部署再上基础模型。顺手还揪出一个测试集数据重复的问题。
一项新研究系统比较了将基础模型 DINOv2 适配到医学影像分割与从头训练 U-Net、ResU-Net 的效果，实验覆盖结肠镜、乳腺超声和皮肤镜三类影像，测试集包含训练来源机构的保留数据和其他机构的独立数据集。
结果显示，适配后的 DINOv2 在每一个外部数据集上都比任务专用模型泛化更好；但在训练分布内，当数据充足时 U-Net 类模型仍有竞争力，且边界勾画更精细，而 DINOv2 更容易漏掉小病灶。研究还发现只需适配基础模型的一小部分参数（从仅调 bias 到全量微调不等），所需容量由成像领域决定而非适配方法。
研究还指出一个广泛使用的息肉测试集存在与另一基准重复的图像。结论是：需要跨机构部署时应选择基础模型适配，并按目标领域选择适配容量。
Medical Image Segmentation: Foundation Model Adaptation versus Task-Specific Training
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 training source and on independent datasets from other institutions. The adapted foundation model generalised better on every external dataset. Within the training distribution, the task-specific models remained competitive when training data were abundant, and delineated boundaries more precisely, while the foundation model more often missed small lesions. Only a small fraction of the foundation model required adaptation, and the capacity needed was determined by the imaging domain rather than by the adaptation method. A widely used polyp test set was also found to duplicate images from another benchmark. Foundation-model adaptation is therefore warranted where models must work across sites, with capacity selected on the target domain.
This report is categorized under AITOP's "Papers" desk. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include DINOv2, 医学影像, 图像分割, 微调.
Source attribution: curated from arXiv cs.AI (@Emre Kutlusan, Yongmin Li).