[论文解读] Radiologist-Level COVID-19 Detection Using CT Scans with Detail-Oriented Capsule Networks

Aryan Mobiny, Pietro Antonio Cicalese+7 more|arXiv (Cornell University)|Apr 16, 2020

COVID-19 diagnosis using AI参考文献 38被引用 73

一句话总结

本文提出 DECAPS，一种具有反向动态路由、看见隐蔽训练（Peekaboo training）以及 pix2pix 数据增强的注重细节的胶囊网络，以从 CT 扫描中检测COVID-19，达到放射科医生水平的性能并优于若干基线模型。

ABSTRACT

Radiographic images offer an alternative method for the rapid screening and monitoring of Coronavirus Disease 2019 (COVID-19) patients. This approach is limited by the shortage of radiology experts who can provide a timely interpretation of these images. Motivated by this challenge, our paper proposes a novel learning architecture, called Detail-Oriented Capsule Networks (DECAPS), for the automatic diagnosis of COVID-19 from Computed Tomography (CT) scans. Our network combines the strength of Capsule Networks with several architecture improvements meant to boost classification accuracies. First, DECAPS uses an Inverted Dynamic Routing mechanism which increases model stability by preventing the passage of information from non-descriptive regions. Second, DECAPS employs a Peekaboo training procedure which uses a two-stage patch crop and drop strategy to encourage the network to generate activation maps for every target concept. The network then uses the activation maps to focus on regions of interest and combines both coarse and fine-grained representations of the data. Finally, we use a data augmentation method based on conditional generative adversarial networks to deal with the issue of data scarcity. Our model achieves 84.3% precision, 91.5% recall, and 96.1% area under the ROC curve, significantly outperforming state-of-the-art methods. We compare the performance of the DECAPS model with three experienced, well-trained thoracic radiologists and show that the architecture significantly outperforms them. While further studies on larger datasets are required to confirm this finding, our results imply that architectures like DECAPS can be used to assist radiologists in the CT scan mediated diagnosis of COVID-19.

研究动机与目标

• 在放射科专业知识受限时，推动快速、准确的 COVID-19 筛查。
• 开发一个能专注于有信息的区域同时利用上下文的模型。
• 通过合成数据增强和有针对性的训练策略来解决数据稀缺问题。
• 与经验丰富的放射科医生进行基准对比，以评估实际临床应用价值。

提出的方法

• 提出 DECAPS，一种胶囊网络架构，包含两大要点：面向细节的路由以突出 ROI 和一种反向动态路由机制。
• 引入 IDR，使低层胶囊相互竞争以吸引高层胶囊的注意力，从而为每个类别产生胶囊头和激活图。
• 实现 Peekaboo 训练 regime，按照激活图裁剪/丢弃补丁，以鼓励从多个 ROI 学习。
• 使用带 PatchGAN 和 U-net 结构的 pix2pix 条件 GAN 进行数据增强，以生成真实的 COVID-19 CT 样本。
• 使用两阶段训练/预测策略，包括来自全图的粗略预测和来自 ROI 的细粒度预测，随后进行蒸馏得到最终预测。

实验结果

研究问题

• RQ1 在数据有限的情况下，基于胶囊的网络能否在基于 CT 的 COVID-19 检测中超越传统的 CNN？
• RQ2 详细的 ROI 集中路由和训练策略是否能提升定位与分类性能？
• RQ3 使用 GAN 生成的 CT 样本进行数据增强对模型泛化有何影响？
• RQ4 该模型在诊断准确性方面与经验丰富的放射科医生相比如何？

主要发现

• 在 Peekaboo 的配合下，DECAPS 超越基线 CNN，并取得最高 AUC 0.961±0.009。
• 单独的 DECAPS 达到 AUC 0.927±0.017，在同一数据集上超过 Inception-v3、DenseNet121 和 ResNet50。
• 结合 DECAPS+Peekaboo 与 pix2pix 增强获得最佳性能（AUC 0.961±0.009）。
• 该模型展示出更高的召回率（0.915±0.057）和相当的精确度（0.843±0.024）相对其他架构。
• 与三位放射科医生相比，该架构在 Alpha-Test 集的性能显著优于人类专家。
• 每个胶囊头的激活图实现对贡献于决策区域的弱监督定位。