arXiv 2606.006022026-05-30cs.CV

ASAP: 基于解剖感知语义自适应预训练的医学体素表示学习

ASAP: Advancing Medical Volumetric Representation Learning with Anatomy-aware Semantically-adaptive Pre-training

School of Biomedical Engineering, Division of Life Sciences and Medicine, University of Science and Technology of China（生物医学工程学院，生命科学与医学系，中国科学技术大学）；Medical Imaging, Robotics, Analytic Computing & Learning (MIRACLE) Lab, YRD-RIGHT, USTC Suzhou Institute for Advanced Research（医学影像、机器人、分析计算与学习（MIRACLE）实验室，YRD-RIGHT，中国科学技术大学苏州研究院）；Jiangsu Provincial Key Laboratory of Multimodal Digital Twin Technology（江苏省多模态数字孪生技术重点实验室）；Biomedical Basic Research Center (BBRC) of Jiangsu Province（江苏省生物医学基础研究中心）；Department of Radiology, The First Affiliated Hospital of USTC, Division of Life Sciences and Medicine, USTC（放射科，中国科学技术大学第一附属医院，生命科学与医学系，中国科学技术大学）；Anhui IFLYTEK CO., Ltd（安徽科大讯飞股份有限公司）；School of Medicine, Stanford University（医学院，斯坦福大学）；State Key Laboratory of Precision and Intelligent Chemistry, Hefei, Anhui, China（安徽省精密与智能化学重点实验室，合肥，安徽，中国）

Rongsheng Wang, Fenghe Tang, Zihang Jiang, Yingtai Li, Xu Zhang, Haoran Lai, Wenxin Ma, Wei Wei, Zhiyang He, Xiaodong Tao, Rui Yan, Qingsong Yao, Shaohua Kevin Zhou

备注：: MICCAI2025 extention

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

提出ASAP框架，通过解剖感知知识注入、语义自适应对齐与融合，从胸部CT扫描和放射学报告中学习可迁移且可解释的体素表示，在15个数据集和22个下游任务上取得最先进性能。

摘要：

从医学体素扫描中学习可迁移和可解释的表示仍然具有挑战性，因为存在复杂的解剖结构和放射学报告提供的弱、异质监督。在本文中，我们提出了解剖感知语义自适应预训练（ASAP），一个用于从大规模胸部CT扫描及其对应放射学报告中进行细粒度医学体素表示学习的原理性视觉-语言预训练框架。ASAP集成了三个关键组件：（1）解剖感知知识注入模块，通过现成的分割工具融入器官级结构先验，以促进解剖上一致的表示；（2）语义自适应选择性对齐机制，动态地将句子级别的发现与局部体素区域关联；（3）语义自适应融合模块，在双模态掩码建模范式下，实现解剖信息视觉特征与基于文本线索之间的有效交互。除了方法论贡献外，我们还为胸部CT上的医学体素视觉-语言预训练建立了一个全面的基准，涵盖15个数据集和22个下游任务，包括异常分类、分割、疾病预后预测、报告生成、词汇分类、跨模态检索和视觉问答。该基准提供了标准化的评估协议，以系统评估在不同临床设置和数据制度下的表示质量。大量实验表明，ASAP在跨任务和数据集上一致地实现了最先进的性能，在有限监督和分布偏移下尤其显著，验证了其在学习可迁移和临床有意义的体素表示方面的有效性。

英文摘要：

Learning transferable and interpretable representations from medical volumetric scans remains challenging due to complex anatomical structures and weak, heterogeneous supervision provided by radiology reports. In this paper, we propose Anatomy-aware Semantically-Adaptive Pre-training (ASAP), a principled vision-language pre-training framework for fine-grained medical volumetric representation learning from large-scale chest CT scans and their corresponding radiology reports. ASAP integrates three key components: (1) an anatomy-aware knowledge injection module that incorporates organ-level structural priors via off-the-shelf segmentation tool to encourage anatomically coherent representations; (2) a semantically-adaptive selective alignment mechanism that dynamically associates sentence-level findings with localized volumetric regions; and (3) a semantically-adaptive fusion module for effective interaction between anatomically informed visual features and grounded textual cues under dual-modal masked modeling paradigm. Beyond methodological contributions, we establish a comprehensive benchmark for medical volumetric vision-language pre-training on chest CT, covering 15 datasets and 22 downstream tasks spanning abnormality classification, segmentation, disease prognosis prediction, report generation, vocabulary classification, cross-modal retrieval and visual question answering. This benchmark provides standardized evaluation protocols to systematically assess representation quality under diverse clinical settings and data regimes. Extensive experiments demonstrate that ASAP consistently achieves state-of-the-art performance across tasks and datasets, with particularly pronounced gains under limited supervision and distribution shift, validating its effectiveness in learning transferable and clinically meaningful volumetric representations.

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。