MaxStyle: Adversarial Style Composition for Robust Medical Image Segmentation
MaxStyle: Adversarial Style Composition for Robust Medical Image Segmentation
复制标题
DOI：
10.48550/arxiv.2206.01737
复制
发表时间：
2022-06
期刊：
影响因子：
--
通讯作者：
Chen Chen-Chen;Zeju Li;C. Ouyang;Matthew Sinclair;Wenjia Bai;D. Rueckert
中科院分区：
文献类型：
--
作者：
Chen Chen-Chen;Zeju Li;C. Ouyang;Matthew Sinclair;Wenjia Bai;D. Rueckert
卷积神经网络（CNN）在训练集和测试集来自同一个域的基准数据集上取得了显着的分割准确性，但它们的性能在看不见的域上会显着下降，这阻碍了CNN在许多临床场景中的部署。大多数现有的工作通过收集多域数据集进行训练来提高模型的域外（OOD）鲁棒性，这是昂贵的，并且由于隐私和后勤问题可能并不总是可行的。在这项工作中，我们专注于仅使用单域数据集来提高模型的鲁棒性。我们提出了一种新的数据增强框架MaxStyle，它最大限度地提高了风格增强模型OOD性能的有效性。它将一个辅助的风格增强图像解码器连接到分割网络，以进行鲁棒的特征学习和数据增强。重要的是，MaxStyle通过使用噪声扩展样式空间并通过对抗训练搜索潜在特征的最差情况样式组成，从而提高了图像样式的多样性和硬度。通过对多个公共心脏和前列腺MR数据集的广泛实验，我们证明了MaxStyle可以显著提高分布外鲁棒性，以防止在低和高训练数据设置下出现不可见的损坏以及多个不同的不可见部位和未知图像序列之间的常见分布偏移。代码可以在https://github.com/cherise215/MaxStyle上找到。
Convolutional neural networks (CNNs) have achieved remarkable segmentation accuracy on benchmark datasets where training and test sets are from the same domain, yet their performance can degrade significantly on unseen domains, which hinders the deployment of CNNs in many clinical scenarios. Most existing works improve model out-of-domain (OOD) robustness by collecting multi-domain datasets for training, which is expensive and may not always be feasible due to privacy and logistical issues. In this work, we focus on improving model robustness using a single-domain dataset only. We propose a novel data augmentation framework called MaxStyle, which maximizes the effectiveness of style augmentation for model OOD performance. It attaches an auxiliary style-augmented image decoder to a segmentation network for robust feature learning and data augmentation. Importantly, MaxStyle augments data with improved image style diversity and hardness, by expanding the style space with noise and searching for the worst-case style composition of latent features via adversarial training. With extensive experiments on multiple public cardiac and prostate MR datasets, we demonstrate that MaxStyle leads to significantly improved out-of-distribution robustness against unseen corruptions as well as common distribution shifts across multiple, different, unseen sites and unknown image sequences under both low- and high-training data settings. The code can be found at https://github.com/cherise215/MaxStyle.