[论文解读] COVID-DA: Deep Domain Adaptation from Typical Pneumonia to COVID-19

Yifan Zhang, Shuaicheng Niu+7 more|arXiv (Cornell University)|Apr 30, 2020

COVID-19 diagnosis using AI参考文献 52被引用 57

一句话总结

COVID-DA 引入一个领域自适应框架，通过区分领域共享与领域专用分类器并对齐特征与联合分布，将知识从典型肺炎迁移到COVID-19 诊断。

ABSTRACT

The outbreak of novel coronavirus disease 2019 (COVID-19) has already infected millions of people and is still rapidly spreading all over the globe. Most COVID-19 patients suffer from lung infection, so one important diagnostic method is to screen chest radiography images, e.g., X-Ray or CT images. However, such examinations are time-consuming and labor-intensive, leading to limited diagnostic efficiency. To solve this issue, AI-based technologies, such as deep learning, have been used recently as effective computer-aided means to improve diagnostic efficiency. However, one practical and critical difficulty is the limited availability of annotated COVID-19 data, due to the prohibitive annotation costs and urgent work of doctors to fight against the pandemic. This makes the learning of deep diagnosis models very challenging. To address this, motivated by that typical pneumonia has similar characteristics with COVID-19 and many pneumonia datasets are publicly available, we propose to conduct domain knowledge adaptation from typical pneumonia to COVID-19. There are two main challenges: 1) the discrepancy of data distributions between domains; 2) the task difference between the diagnosis of typical pneumonia and COVID-19. To address them, we propose a new deep domain adaptation method for COVID-19 diagnosis, namely COVID-DA. Specifically, we alleviate the domain discrepancy via feature adversarial adaptation and handle the task difference issue via a novel classifier separation scheme. In this way, COVID-DA is able to diagnose COVID-19 effectively with only a small number of COVID-19 annotations. Extensive experiments verify the effectiveness of COVID-DA and its great potential for real-world applications.

研究动机与目标

• 在 COVID-19 标注稀缺时，推动从胸部放射影像诊断COVID-19。
• 同时解决肺炎与COVID-19之间的域分布不匹配和任务差异。
• 开发一个深度学习模型，利用带标签的肺炎数据在目标标签有限的情况下提升COVID-19诊断。

提出的方法

• 使用一个域共享的特征提取器来学习域不变的表示。
• 结合两个域判别器以对齐特征分布和联合特征-预测分布。
• 引入一个分类器分离方案，包含一个域共享分类器和两个域专用分类器（每个域各一个）。
• 最大化域共享分类器与域专用分类器之间的多样性，以捕捉域私有信息。
• 使用 focal loss 进行训练，以处理标签不平衡并提升判别能力。

实验结果

研究问题

• RQ1 域对抗学习是否能够在尽管存在域偏移的情况下对齐肺炎与COVID-19 图像的特征？
• RQ2 将分类器分离为共享组件和域专用组件是否能在目标标签有限的情况下改善 COVID-19 诊断？
• RQ3 显式最大化分类器之间的多样性对跨域诊断性能有何影响？
• RQ4 与基线和其他领域自适应方法相比，COVID-DA 在 COVID-19 诊断上的表现如何？
• RQ5 方法对医学影像数据集中常见的类不平衡是否具有鲁棒性？

主要发现

• COVID-DA 在评估方法中在 COVID-19 诊断上实现最高的 F1 为 92.98% 和 AUC 为 0.985。
• COVID-DA 优于 Source-only、Target-only、Fine-tuning、标准 DA 方法（MCD、DANN、DSN、DMAN），以及半监督 DA 方法（SDT、semi-DMAN）。
• 消融研究表明所有组件（特征对抗适应、分类器对抗适应、分类器多样性、焦点损失）对性能提升均有贡献。
• Visual Grad-CAM 分析表明域共享和目标特定分类器聚焦于诊断的互补区域。
• 域共享与域专用分类器的集成提升了鲁棒性与临床可解释性。