arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2602.23916cs.CVcs.AI
收藏

基于拓扑驱动的医学基础模型分割迁移性估计

Topology-Driven Transferability Estimation of Medical Foundation Models for Segmentation

Peking University(北京大学);Hohai University(河海大学);Beijing Normal University-Hong Kong Baptist University United International College(北京师范大学-香港 Baptist大学联合国际学院);National Institute of Health Data Science, Peking University(健康数据科学国家研究院,北京大学);Institute of Medical Technology, Peking University(北京大学医学技术研究院);State Key Laboratory of General Artificial Intelligence, Peking University(通用人工智能国家重点实验室,北京大学)

Jiaqi Tang, Shaoyang Zhang, Xiaoqi Wang, Jiaying Zhou, Yang Liu, Qingchao Chen

AI总结:

提出拓扑驱动迁移性估计框架,通过全局表示拓扑散度、局部边界感知拓扑一致性和任务自适应融合,无需微调即可高效选择医学基础模型,在OpenMind基准上加权Kendall指标相对提升约31%。

摘要:

大规模自监督学习(SSL)的出现产生了大量的医学基础模型。然而,为特定分割任务选择最优的医学基础模型仍然是一个计算瓶颈。现有的迁移性估计(TE)指标主要针对分类任务设计,依赖于全局统计假设,无法捕捉密集预测所需的拓扑复杂性。我们提出了一种新颖的拓扑驱动迁移性估计框架,评估流形可处理性而非统计重叠。我们的方法引入了三个组成部分:(1)全局表示拓扑散度(GRTD),利用最小生成树量化特征-标签结构同构性;(2)局部边界感知拓扑一致性(LBTC),专门在关键解剖边界评估流形可分离性;(3)任务自适应融合,根据目标任务的语义基数动态整合全局和局部指标。在跨不同解剖目标和SSL基础模型的大规模OpenMind基准上验证,我们的方法在加权Kendall指标上显著优于最先进的基线,相对提升约31%,提供了一种鲁棒的、无需训练的代理,用于高效模型选择而无需微调成本。代码将在接收后公开。

英文摘要:

The advent of large-scale self-supervised learning (SSL) has produced a vast zoo of medical foundation models. However, selecting optimal medical foundation models for specific segmentation tasks remains a computational bottleneck. Existing Transferability Estimation (TE) metrics, primarily designed for classification, rely on global statistical assumptions and fail to capture the topological complexity essential for dense prediction. We propose a novel Topology-Driven Transferability Estimation framework that evaluates manifold tractability rather than statistical overlap. Our approach introduces three components: (1) Global Representation Topology Divergence (GRTD), utilizing Minimum Spanning Trees to quantify feature-label structural isomorphism; (2) Local Boundary-Aware Topological Consistency (LBTC), which assesses manifold separability specifically at critical anatomical boundaries; and (3) Task-Adaptive Fusion, which dynamically integrates global and local metrics based on the semantic cardinality of the target task. Validated on the large-scale OpenMind benchmark across diverse anatomical targets and SSL foundation models, our approach significantly outperforms state-of-the-art baselines by around 31% relative improvement in the weighted Kendall metric, providing a robust, training-free proxy for efficient model selection without the cost of fine-tuning. The code will be made publicly available upon acceptance.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑