arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2201.10737cs.CVcs.AIcs.LGeess.IV
收藏

基于类感知的对抗性Transformers医学图像分割

Class-Aware Adversarial Transformers for Medical Image Segmentation

Yale University(耶鲁大学);UT Austin(得克萨斯大学奥斯汀分校);University of Oxford(牛津大学)

Chenyu You, Ruihan Zhao, Fenglin Liu, Siyuan Dong, Sandeep Chinchali, Ufuk Topcu, Lawrence Staib, James S. Duncan

摘要:

Transformers已经在医学图像分析领域内对远程依赖性进行建模方面取得了显著的进展。然而,现有的基于变换器的模型存在以下缺点:(1)由于标记化方案过于简单,无法捕捉图像的重要特征;(2)由于只考虑单尺度特征表示,模型存在信息损失;以及(3)在不考虑丰富的语义上下文和解剖纹理的情况下,由模型生成的分割标签图不够准确。在这项工作中,我们提出了CASTformer,一种新型的对抗Transformers,用于2D医学图像分割。首先,我们利用金字塔结构来构建多尺度表示和处理多尺度变化。然后,我们设计了一个新的类感知的Transformer模块,以更好地学习具有语义结构的对象的区分区域。最后,我们利用对抗性训练策略来提高分割精度,并相应地允许基于transformer的神经网络捕获高级语义相关内容和低级解剖特征。我们的实验表明,CASTformer显着优于以前的国家的最先进的变压器为基础的方法在三个基准,获得2.54%-5.88%的绝对改善骰子比以前的模型。进一步的定性实验提供了模型内部工作的更详细的图片,揭示了提高透明度的挑战,并证明了迁移学习可以大大提高性能并减少训练中医学图像数据集的大小,使CASTformer成为下游医学图像分析任务的有力起点。

英文摘要:

Transformers have made remarkable progress towards modeling long-range dependencies within the medical image analysis domain. However, current transformer-based models suffer from several disadvantages: (1) existing methods fail to capture the important features of the images due to the naive tokenization scheme; (2) the models suffer from information loss because they only consider single-scale feature representations; and (3) the segmentation label maps generated by the models are not accurate enough without considering rich semantic contexts and anatomical textures. In this work, we present CASTformer, a novel type of adversarial transformers, for 2D medical image segmentation. First, we take advantage of the pyramid structure to construct multi-scale representations and handle multi-scale variations. We then design a novel class-aware transformer module to better learn the discriminative regions of objects with semantic structures. Lastly, we utilize an adversarial training strategy that boosts segmentation accuracy and correspondingly allows a transformer-based discriminator to capture high-level semantically correlated contents and low-level anatomical features. Our experiments demonstrate that CASTformer dramatically outperforms previous state-of-the-art transformer-based approaches on three benchmarks, obtaining 2.54%-5.88% absolute improvements in Dice over previous models. Further qualitative experiments provide a more detailed picture of the model's inner workings, shed light on the challenges in improved transparency, and demonstrate that transfer learning can greatly improve performance and reduce the size of medical image datasets in training, making CASTformer a strong starting point for downstream medical image analysis tasks.

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。

↑