arXiv 2309.118762025-02-13cs.CVcs.AI

用于体医学图像分割预训练的多级非对称对比学习

Multi-level Asymmetric Contrastive Learning for Volumetric Medical Image Segmentation Pre-training

Peking University（北京大学）；Peking University Health Science Center（北京大学医学部）；National Biomedical Imaging Center（国家生物医学成像中心）；Peking University Shenzhen Graduate School（北京大学深圳研究生院）；Georgia Institute of Technology（佐治亚理工学院）；UT Southwestern Medical Center（西南大学达拉斯医学中心）

Shuang Zeng, Lei Zhu, Xinliang Zhang, Micky C Nnamdi, Wenqi Shi, J Ben Tamo, Qian Chen, Hangzhou He, Lujia Jin, Zifeng Tian, Qiushi Ren, Zhaoheng Xie, Yanye Lu

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

针对医学图像分割标注数据匮乏问题，提出MACL多级非对称对比学习框架，通过同时预训练编码器和解码器、整合多级表示，提升分割性能与泛化能力，在多个数据集上优于现有策略。

摘要：

医学图像分割是一项基础且具有挑战性的任务，因为从专家处获取大量高质量标注数据的过程十分艰巨。对比学习为这一困境提供了有前景但仍存在问题的解决方案。首先，现有医学对比学习策略侧重于提取图像级表示，忽略了丰富的多级表示。此外，它们要么通过随机初始化，要么通过与编码器分开预训练来未充分利用解码器，从而忽视了编码器和解码器之间潜在的协作。为解决这些问题，我们提出了一种名为MACL的新型多级非对称对比学习框架，用于体医学图像分割预训练。具体而言，我们设计了一种非对称对比学习结构，同时预训练编码器和解码器，为分割模型提供更好的初始化。此外，我们开发了一种多级对比学习策略，整合特征级、图像级和像素级表示之间的对应关系，确保编码器和解码器在预训练阶段从不同尺度和粒度的表示中捕获全面细节。最后，在8个医学图像数据集上的实验表明，我们的MACL框架优于现有的11种对比学习策略。例如，在10%标注数据下，MACL在ACDC、MMWHS、HVSMR和CHAOS数据集上的Dice分别比之前的最佳结果高1.72%、7.87%、2.49%和1.48%，且可视化图显示其预测更精确。此外，MACL在5种变体U-Net骨干网络中具有很强的泛化能力。我们的代码将在https://github.com/stevezs315/MACL发布。

英文摘要：

Medical image segmentation is a fundamental yet challenging task due to the arduous process of acquiring large volumes of high-quality labeled data from experts. Contrastive learning offers a promising but still problematic solution to this dilemma. Firstly existing medical contrastive learning strategies focus on extracting image-level representation, which ignores abundant multi-level representations. Furthermore they underutilize the decoder either by random initialization or separate pre-training from the encoder, thereby neglecting the potential collaboration between the encoder and decoder. To address these issues, we propose a novel multi-level asymmetric contrastive learning framework named MACL for volumetric medical image segmentation pre-training. Specifically, we design an asymmetric contrastive learning structure to pre-train encoder and decoder simultaneously to provide better initialization for segmentation models. Moreover, we develop a multi-level contrastive learning strategy that integrates correspondences across feature-level, image-level, and pixel-level representations to ensure the encoder and decoder capture comprehensive details from representations of varying scales and granularities during the pre-training phase. Finally, experiments on 8 medical image datasets indicate our MACL framework outperforms existing 11 contrastive learning strategies. i.e. Our MACL achieves a superior performance with more precise predictions from visualization figures and 1.72%, 7.87%, 2.49% and 1.48% Dice higher than previous best results on ACDC, MMWHS, HVSMR and CHAOS with 10% labeled data, respectively. And our MACL also has a strong generalization ability among 5 variant U-Net backbones. Our code will be released at https://github.com/stevezs315/MACL.

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。