从掩码引导自监督中学习可泛化的 3D 医学图像表征
Learning Generalizable 3D Medical Image Representations from Mask-Guided Self-Supervision
把自动生成的类无关掩码当作上下文分割的预训练 pretext task，为 3D 医学体数据提供一种无需标注、可迁移到少样本分割与冻结编码器分类的预训练范式；其「掩码提议即结构监督」的思路也可迁移到自然图像或视频的 3D/时序表征预训练。
原文依据 treats in-context segmentation as the pretext task for learning general-purpose medical imaging representations
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论基础模型通过从大规模无标注数据中学习通用表征，已经改变了视觉与语言领域，然而 3D 医学影像仍缺乏类似的方法。现有的自监督方法依赖低层重建或对比目标，无法捕捉医学图像分析所关键的解剖语义，从而限制了对下游任务的迁移。我们提出 MASS（MAsk-guided Self-Supervised learning，掩码引导自监督学习），将上下文内分割作为学习通用医学影像表征的预训练任务。MASS 的核心洞见是：自动生成的类无关掩码能够为学习语义丰富的表征提供充分的结构监督。通过在涵盖解剖结构与病理发现的数千个多样化掩码提议上进行训练，MASS 学到了医学结构的语义定义要素：外观、形状、空间上下文与解剖关系的整体组合。我们在多种数据规模下验证了其有效性：从在单个数据集上的小规模预训练（20–200 例扫描），到在 5K 例 CT、MRI 与 PET 体数据上进行大规模多模态预训练，全程无需标注。MASS 展示了：(i) 对新结构的少样本分割；(ii) 仅用 20–40% 的标注数据即可匹配全监督，并在低数据情形下以超过 20 的 Dice 分数优于自监督基线；(iii) 在未见病理上的冻结编码器分类，可匹配使用数千样本的全监督训练。掩码引导的自监督预训练捕捉到了广泛可泛化的知识，为无需专家标注的 3D 医学影像基础模型开辟了一条路径。代码见 https://github.com/Stanford-AIMI/MASS。
英文原文摘要 1791 字符
Foundation models have transformed vision and language by learning general-purpose representations from large-scale unlabeled data, yet 3D medical imaging lacks analogous approaches. Existing self-supervised methods rely on low-level reconstruction or contrastive objectives that fail to capture the anatomical semantics critical for medical image analysis, limiting transfer to downstream tasks. We present MASS (MAsk-guided Self-Supervised learning), which treats in-context segmentation as the pretext task for learning general-purpose medical imaging representations. MASS's key insight is that automatically generated class-agnostic masks provide sufficient structural supervision for learning semantically rich representations. By training on thousands of diverse mask proposals spanning anatomical structures and pathological findings, MASS learns what semantically defines medical structures: the holistic combination of appearance, shape, spatial context, and anatomical relationships. We demonstrate effectiveness across data regimes: from small-scale pretraining on individual datasets (20-200 scans) to large-scale multi-modal pretraining on 5K CT, MRI, and PET volumes, all without annotations. MASS demonstrates: (i) few-shot segmentation on novel structures, (ii) matching full supervision with only 20-40% labeled data while outperforming self-supervised baselines by over 20 in Dice score in low-data regimes, and (iii) frozen-encoder classification on unseen pathologies that matches full supervised training with thousands of samples. Mask-guided self-supervised pretraining captures broadly generalizable knowledge, opening a path toward 3D medical imaging foundation models without expert annotations. Code is available \href https://github.com/Stanford-AIMI/MASS here .