对比九种自监督学习方法在多模态医学图像分割中的迁移性能，发现一种新方法效果最优。
Benchmarking transferability of SSL pretraining to same and different modality segmentation tasks
- 用九种自监督学习方法在1万张CT影像上预训练，再迁移到九个分割任务中测试。
- 基于掩码建模和自蒸馏的SMIT方法表现最佳，少样本时提升最明显，收敛最快。
- 小样本下预训练方法差异大，适合标注数据稀缺场景，如肿瘤分割等临床应用。
本文使用相同10,412例3D CT扫描数据（共189万张2D轴向切片）覆盖多种疾病部位，对九种涵盖四类预训练任务的自监督学习方法从头开始进行预训练。将每种方法的Swin Transformer编码器嵌入SwinUNETR结构（3D CNN解码器+跳连），在九个不同复杂度的公开分割任务上微调，包括腹部大器官、头颈部结构及CT/MRI中的肿瘤。使用Dice相似系数（DSC）评估性能。进一步通过中心核对齐分析微调收敛速度、跨模态迁移（CT到MRI）能力以及少样本与多样本微调下的特征复用模式。结果表明：结合掩码图像建模（MIM）与局部/全局自蒸馏的自蒸馏掩码图像变换器（SMIT）在全部九项任务中取得最高平均分割精度，微调收敛最快，少样本到多样本性能差距最小，体现最强数据效率；其特征复用模式在少样本与多样本间最为一致。基于MIM的方法（SimMIM）及自蒸馏方法（DINO, iBOT）优于依赖图像级全局表示的对比学习与旋转预测方法。不同自监督方法间的性能差异在少样本设置下最大，随着标注微调数据量增加而缩小，说明在标注预算有限时，预训练方法的选择至关重要。
原文摘要 · Abstract (English)
Methods: Nine SSL methods spanning four pretext-task families were pretrained from scratch using the same 10{,}412 3D CT scans (1.89~M 2D axial slices) covering varied disease sites. The pretrained Swin Transformer encoder from each method was integrated into a SwinUNETR-style segmentation network (Swin encoder with a 3D CNN decoder and skip connections) and fine-tuned on nine public segmentation tasks of varying complexity, including large abdominal organs, head-and-neck structures, and tumors from CT and MRI. Performance was assessed using Dice similarity coefficient (DSC). Fine-tuning convergence speed, transferability across modalities (CT-to-MRI), and feature-reuse patterns between few- and many-shot fine tuning were further analyzed using centered kernel alignment. Results: Self-distilled masked image transformer (SMIT), which combines masked image modeling (MIM) with local and global self-distillation, achieved the highest overall segmentation accuracy across the nine tasks, the fastest fine-tuning convergence, and the smallest few-shot-to-many-shot performance gap, indicating the strongest data efficiency. SMIT also showed the most consistent feature-reuse patterns between few- and many-shot fine tuning. MIM-based SimMIM and self-distillation methods (DINO, iBOT) outperformed contrastive learning and rotation prediction, which rely on image-level global representations. Differences between SSL methods were largest in the few-shot setting and narrowed as the size of the labeled fine-tuning dataset increased, indicating that the choice of SSL pretraining matters most under limited annotation budgets.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。