发表机构
Universidad de Chile; King Abdullah University of Science and Technology (KAUST); Universidad de los Andes（智利大学; 阿卜杜拉国王科技大学; 安第斯大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出B-MIM方法改进iBOT目标，通过降低全局语义压力增强编码器对细粒度解剖结构的特征捕获，在跨数据集肝血管、肿瘤分割任务中仅更新部分参数就实现了更优性能，提升了模型泛化能力。
AI 中文摘要
自监督预训练可为医学成像提供可迁移的表示，但大多数CT编码器仍偏向于粗粒度语义理解，限制了其对血管或小肿瘤等细粒度解剖结构的敏感性。本文提出偏置掩码图像建模（Biased Masked Image Modeling，B-MIM），这是对iBOT目标的改进，该方法会随机降低全局语义对齐，以优先考虑局部块重建。这种偏置鼓励编码器捕获高频形态细节和结构连续性。我们整理了来自17个公开来源的9955项经筛选研究的多机构CT腹部数据集，并使用B-MIM预训练了一个3D Swin Transformer骨干网络。在跨数据集的肝血管分割实验中，与完全微调的基线相比，所提出的编码器仅更新部分参数，就提高了拓扑保真度（clDice）并在肿瘤分割中取得了有竞争力的Dice分数。我们的结果表明，预训练期间降低全局语义压力可增强对复杂解剖结构的泛化能力。
英文摘要
Self-supervised pretraining enables transferable representations for medical imaging, yet most CT encoders remain biased toward coarse semantic understanding, limiting their sensitivity to fine-grained anatomical structures such as vessels or small tumors. In this paper, we introduce Biased Masked Image Modeling (B-MIM), a modification of the iBOT objective that stochastically reduces global semantic alignment to prioritize local patch reconstruction. This bias encourages the encoder to capture high-frequency morphological details and structural continuity. We curate a multi-institutional CT abdominal dataset of 9,955 filtered studies from 17 public sources and pretrain a 3D Swin Transformer backbone using B-MIM. Across inter-dataset experiments on liver vessel segmentation, the proposed encoder improves topological fidelity (clDice) and achieves competitive Dice scores in tumor segmentation, compared to fully fine-tuned baselines, despite updating only a fraction of the parameters. Our results suggest that reducing global semantic pressure during pretraining enhances generalization to intricate anatomical structures.
CommentsPublished at MedAGI in MICCAI 2026
