Print

发布时间: 2020-10-16
摘要点击次数:
全文下载次数:
DOI: 10.11834/jig.200222
2020 | Volume 25 | Number 10




    前沿进展    




  <<上一篇 




  下一篇>> 




新型冠状病毒肺炎(COVID-19)医学影像AI诊断研究进展
expand article info 孟琭, 李镕辉
东北大学信息科学与工程学院, 沈阳 110004
收稿日期: 2020-05-29; 修回日期: 2020-07-11; 预印本日期: 2020-07-18
基金项目: 国家自然科学基金项目（61973058）；中央高校基本科研业务费专项资金项目（N2004020）
第一作者简介: 孟琭, 1982年生, 男, 副教授, 主要研究方向为人工智能、图像处理、计算机辅助诊断、区块链。E-mail:menglu@ise.neu.edu.cn.
中图法分类号: TP
文献标识码: A
文章编号: 1006-8961(2020)10-2058-10

摘要

2020年3月，世界卫生组织（World Health Organization，WHO）宣布新型冠状病毒肺炎（corona virus disease 2019，COVID-19）为世界大流行病，疫情的爆发给世界各地医疗系统带来巨大压力。现有的COVID-19诊断标准是核酸检测阳性，然而核酸检测假阴性率高达17%~25.5%，为避免漏诊，需要采用基于影像学的AI诊断方法筛查大量疑似病例，扼制疾病传播。本综述将回顾疫情爆发数月以来，基于医学影像的新冠肺炎AI辅助诊断的研究成果。首先介绍CT（computed tomography）和X光片的优缺点，以及COVID-19的放射学特征，然后对数据准备、图像分割和分类识别等AI诊断的关键步骤分别进行阐述，最后介绍COVID-19的跟踪和预后（预先对疾病后续发展过程及结果的判断和估计）。本文还整理了部分公开的COVID-19相关数据集，并对数据标注不足的问题提供了弱监督学习和迁移学习等解决方案。实验验证，AI系统诊断COVID-19的敏感性达到97.4%，特异性达到92.2%，优于放射科医生的诊断结果。其中表现尤为突出的是基于语义分割网络检测COVID-19感染区域，由此可以定量分析感染率。AI系统可以辅助医生诊断和治疗COVID-19，提高放射科医生阅读X光片和CT的效率。

关键词

人工智能; 新型冠状病毒肺炎(COVID-19); 图像分割; 计算机辅助诊断; 感染区域分割

Progress of artificial intelligence diagnosis and prognosis technology for COVID-19 medical imaging
expand article info Meng Lu, Li Ronghui
College of Information Science and Engineering, Northeastern University, Shenyang 110004, China
Supported by: National Natural Science Foundation of China (61973058); Fundamental Research Funds for the Central Universities (N2004020)

Abstract

In March 2020, the World Health Organization(WHO) declared the new corona virus pneumonia (COVID-19) as a world pandemic, which means that the epidemic has broken out worldwide. The outbreak of COVID-19 threatens the lives and property safety of countless people and brings great pressure to medical systems. The main clinical symptoms of COVID-19 are fever, cough, and fatigue, which may lead to a fatal complication: acute respiratory distress syndrome. The main challenge in inhibiting the spread of this disease is the lack of efficient detection methods. Although reverse transcription-polymerase chain reaction (RT-PCR) is the gold standard for confirming COVID-19, it takes 4-6 h to obtain the results, and the false-negative rate of RT-PCR detection is as high as 17%-25.5%. Therefore, multiple RT-PCR detections at intervals of several days must be performed to confirm the diagnostic result. In addition, RT-PCR reagents are lacking in many severe epidemic areas. By contrast, X-ray and CT(computed tomography) examination equipment have been widely popularized in hospitals. In clinical practice, by combining clinical symptoms and travel history, CT is an efficient and safe method to diagnose COVID-19. Compared with CT, X-ray examination has faster scanning speed and lower radiation amount. Moreover, X-ray and CT images are important tools for doctors to track and observe the condition and evaluate the efficacy. In summary, medical imaging plays a vital role in limiting the spread of viruses and treating COVID-19. During the outbreak of the epidemic, medical imaging-based AI-assisted diagnostic technology has become a popular research direction. Computer-aided diagnostic technology improves the sensitivity and specificity of doctors' diagnosis and is accurate and efficient, which helps rapid diagnosis of a large number of suspicious cases. For example, the out preformed AI-assisted diagnosis system can achieve an accuracy rate comparable to that of radiologists, and it take less than 1 second to perform a diagnosis. The system has been used in 16 hospitals, with more than 1 300 diagnoses performed daily. This article reviews the latest research works on AI-assisted diagnosis of COVID-19 and analyzes and summarizes them on the basis of four aspects: data preparation, image segmentation, diagnosis, and prognosis. First, this article organizes some public data sets to support the AI-assisted diagnostic technology of COVID-19 and provides several solutions to insufficient datasets, such as the human-in-the-loop strategy, which improves the efficiency of data set production. Using transfer learning, weakly supervised or unsupervised learning can reduce model's dependence on the COVID-19 dataset. Second, the semantic segmentation network is also an indispensable part of the intelligent diagnosis of COVID-19. Segmenting the lung region from the original image is a key pre-processing step, which can reduce the calculation amount of subsequent algorithms. The lesion area helps the doctor to track the condition of the disease, and the infection rate can be calculated according to the size of the infected area. U-Net, U-Net++, and attention U-Net are suitable for the segmentation of medical images because of the small number of parameters, which is not easy to overfit. Furthermore, training the semantic segmentation network with the idea of the generative adversarial network (GAN) can improve the Dice coefficient. Third, this article introduces the AI diagnostic system from two aspects of CT images and X-rays. Comparing different diagnostic schemes, the method of diagnosis based on the segmentation images is better than that based on the original images. Among the classification networks, ResNet and VGG19(visual geometry group 19-layer net) perform better. Methods such as GAN, location attention mechanisms, transfer learning, and combining 2D and 3D features can be used to improve accuracy. In addition, clinical information (travel and contact history, white blood cell count, fever, cough, sputum, patient age, and patient gender) can be used as a basis for diagnosis. For example, algorithm D_FF_Conic uses clinical information as a diagnostic basis and has reached an accuracy rate of 90%. Clinicians will consider medical imaging and clinical information in the process of diagnosis, but the current AI diagnostic system cannot integrate multiple types of data for diagnosis. Although some algorithms can fuse the diagnostic results of medical images with the diagnostic results of clinical information, the simple fine-tuned algorithm haven't learned the deep internal connection between different types of data. Fourth, AI technology can also predict high-risk patients on the basis of infection rates and clinical information. Some research predicted the survival rate of COVID-19 patients on the basis of age, syndrome, and infection rate. Such algorithms can help doctors find and treat high-risk patients early, thereby reducing mortality, which is of great significance. This article shows the latest progress of COVID-19's medical imaging-based AI diagnosis. Although some AI-assisted diagnostic systems have been deployed in hospitals to play a practical role, these algorithms still have some problems, such as insufficient training data, a single diagnostic basis, and the ability to distinguish between non-COVID-19 pneumonia and COVID-19.

Key words

artificial intelligence; COVID-19; image segmentation; computer aided diagnosis; infection region segmentation

0 引言

2019年12月，中国武汉爆发了新型冠状病毒(Huang等，2020；Lu等，2020)，这种冠状病毒感染的传染病被世界卫生组织命名为2019年冠状病毒病(corona virus disease 2019, COVID-19)(WHO，2020a)。新型冠状病毒肺炎主要的临床症状为发热、咳嗽和乏力(国家卫生健康委员会办公厅和国家中医药管理局办公室，2020)，可能会引发致命的急性呼吸窘迫综合征(acute respiratory distress syndrome, ARDS)(Chen等，2020b)。

控制这种疾病传播的主要障碍是缺乏高效的检测方法。尽管逆转录聚合酶链式反应(reverse transcription-polymerase chain reaction, RT-PCR)是确认COVID-19的金标准(Ai等，2020)，但获得结果需要4~6个小时，且RT-PCR测试还具有较高的假阴性率，需要间隔几天多次进行RT-PCR检测才能确定诊断结果(国家卫生健康委员会办公厅和国家中医药管理局办公室，2020)。此外，在很多疫情严重地区RT-PCR试剂也严重不足。相比之下，X射线和CT(computed tomography)检查设备在各大医院已广泛普及。在临床实践中，通过结合临床症状和体征，CT是识别COVID-19的更高效、更安全的方法。与CT相比，X射线检查具有扫描速度更快、辐射量更低的优点。同时，X射线和CT图像还可应用在COVID-19的跟踪和预后：轻型和普通型患者的肺部影像表现为肺部的多发性小斑片影及间质改变，以肺外带明显; 重型和危重型患者，多发展为双肺多发性的毛玻璃表现与浸润影，重者可出现肺实变(国家卫生健康委员会办公厅和国家中医药管理局办公室，2020)。综上所述，医学影像在限制病毒传播以及治疗COVID-19的过程中起着至关重要的作用。

在COVID-19的医学影像诊断方面，由于需要筛查和治疗大量的病患，为减轻放射科医生的阅片工作量，可以用人工智能技术对医学影像进行自动诊断，计算感染严重程度。Jin等人(2020)设计了一个AI辅助诊断系统，可以实现与放射科医生相媲美的性能，该系统已被部署到16家医院。此外，AI技术还可以根据COVID-19的临床特征，预测高危病患，帮助医生及早发现并治疗，降低死亡率(Yan等，2020)。本文重点关注基于医学影像的AI辅助诊断，从数据准备、图像分割、分类、跟踪和预后方面分别进行阐述。最后提出了几个尚未解决的问题和挑战。

1 数据准备

1.1 公开的数据集

训练深度学习网络模型需要有大规模高质量的数据集，由于迫切需要数据集来支持人工智能技术进行COVID-19的辅助诊断, 部分公开的数据集见表 1。

表 1 公开的数据集
Table 1 Public dataset

下载CSV
数据集名称 类型 组成 描述 文献
COVID-19 CT segmentation dataset CT COVID-19 COVID-19感染区域分割数据集 MedSeg(2020)
JSRT Dataset X光片 正常 肺部区域分割数据集 Shiraishi等人(2000)
COVID-19 BSTI Imaging Database CT COVID-19 帮助医生了解COVID-19影像 BSTI(2020)
COVID-CT-Dataset CT COVID-19其他 349幅COVID-19的CT影像 He等人(2020)
covid-chestxray-dataset X光片，CT COVID-19非COVID-19肺炎健康 一个可用于计算机分析COVID-19的大型X和CT数据集 Cohen等人(2020)
COVIDx X光片 COVID-19非COVID-19肺炎正常 13 975幅X光片 Wang和Wong(2020a)
Open source ultrasound (POCUS) data collection initiative for COVID-19 超声波数据 COVID-19非COVID-19肺炎正常 COVID-19超声波检测数据集 Born等人(2020)
COVID-19 Chest X-ray Database X光片 COVID-19非COVID-19肺炎正常 2 905幅X光片 Chowdhury等人(2020)
Augmented COVID-19 X-ray Images Dataset X光片 COVID-19其他 COVID-19和非COVID-19的数据增强 Alqudah等人(2020a)
Chest X-Ray Images (Pneumonia) X光片 细菌/病毒性肺炎、正常 有5 863幅X射线图像 Kermany等人(2018)
COVID-19 Open Research Dataset Challenge (CORD-19) 论文 文本挖掘数据集 Allen Institute for AI等人(2020a)
COVID-19 Open Research Dataset (CORD-19) 论文 文本挖掘数据集 Allen Institute for AI等人(2020b)
LITCOVID 论文 根据不同的主题和地区做了分类 Chen等人(2020c)
Global research on COVID-19 论文 WHO全球研究数据 WHO(2020b)

尽管已经有一些COVID-19的公开数据集，但数据量与训练神经网络所需的相比还是很缺乏，可以综合使用上面几个数据集，如COVIDx整合了3个开源数据集构成，并补充了JSRT Dataset、Chest X-Ray Images (Pneumonia)这些非COVID-19的肺部数据集。列出的论文数据集，一方面可以帮助学者跟踪COVID-19研究进展；另一方面可以作为文本挖掘数据集，帮助医学界寻找一些问题的答案。

1.2 缓解数据不足问题

COVID-19感染区域分割数据集匮乏的原因之一是标注数据是一项劳动密集型工作——放射科医生手动标注出CT影像的感染区域需要1—5 h。为了提高放射科医生的标注效率，Shan等人(2020)提出了“人工在环(human-in-the-loop, HITL)”策略和VB-Net语义分割网络模型，HITL策略让放射科医生先手动标注一小部分CT图像以初步训练VB-Net，然后用VB-Net自动分割感染区域，经医生校正后作为新的数据再次训练VB-Net，操作流程如图 1。自动分割出的结果和手动分割相比，Dice相似系数为91.6 % ±10.0 %；HITL策略可以将标注时间大幅减少到4 min，体现了医生与算法专家深度合作的优势，这种策略也可以应用在其他图像分割数据集的制作上。



图 1 “人工在环”工作流程(Shan等，2020)
Fig. 1 The human-in-the-loop workflow(Shan et al., 2020)

尽管COVID-19的数据集很缺乏，但有大量其他数据集可以进行迁移学习。Zhang等人(2020b)提出一种深度领域自适应法，进行普通肺炎到COVID-19的领域适应，采用ResNet18模型达到了98.2 %的精确率和88.33 %的敏感性。迁移学习可以将有大量标签的源域知识迁移到只有少量标签或无标签的目标域，同时利用开源的预训练模型缩短训练时间。值得注意的是，弱监督或无监督学习也可缓解数据不足问题，如Zheng等人(2020)训练了一个弱监督二分类网络，该模型仅需要每位患者是否为阳性的标签。具体工作如下：训练一个无监督学习网络，对肺部区域进行分割，然后用提出的DeCoVNet判断是否为COVID-19阳性。

2 图像语义分割

在COVID-19的AI辅助诊疗中，从原始影像中分割出肺部区域是一项关键的预处理步骤，可以减少后续算法的计算量；而分割出病变区域，有助于医生跟踪观察病情，同时可根据感染区域的大小计算出感染率。由于医学影像数据集的限制，相对于DeepLab和DenseNets等语义分割模型而言，U-Net参数量更少，不容易过拟合，因此U-Net及其变体网络(3D U-Net、U-Net+ +和Attention U-Net等)被广泛应用于医学图像的分割。Zheng等人(2020)采用U-Net模型进行肺部区域分割；V-Net作为U-Net的一种3D变体，Butt等人(2020)使用V-Net模型进行COVID-19感染区域的分割。

Gaál等人(2020)采用生成对抗网络和Attention U-Net设计了肺部区域分割网络，在JSRT(Japanese society of radiological technology)数据集(Shiraishi等，2020)上实现97.5 %的Dice。Alom等人(2020)使用NABLA-N(Alom等，2019)网络模型分割CT和X光片中的肺部受到COVID-19感染的区域，效果如图 2。



图 2 正常和COVID-19病例的X射线和CT图像，NABLA-3网络的输入和输出(Alom等，2020)
Fig. 2 X-ray and CT images of normal and COVID-19 cases, inputs and outputs of NABLA-3 network((a) normal images; (b)COVID-19 images; (c)input of NABLA-3; (d)output of NABLA-3 (Alom et al., 2020)

3 医学影像的AI诊断

3.1 基于CT的COVID-19诊断

CT在诊断COVID-19、监测疾病进展和评估疗效等方面具有重要意义(Ai等，2020)。基于CT图像的COVID-19诊断工作有Zheng等人(2020)、Gozes等人(2020)、Shan等人(2020)、Li等人(2020a，b)、Butt等人(2020)、Alom等人(2020)、Maghdid等人(2020a)、Hu等人(2020)、Jin等人(2020)、Chen等人(2020a)、Wang等人(2020b)、Zhang等人(2020a)以及Bai等人(2020)。这些算法可分为二分类和三分类，二分类算法的分类结果为COVID-19阳性和阴性，三分类算法的分类结果为健康、COVID-19阳性和非COVID-19肺炎。

3.1.1 二分类

相对于多元分类，二元分类复杂度低，不确定性小。尽管二分类算法没有充分考虑非COVID-19肺炎的诊断，但是二分类可以通过平移决策边界提高敏感性。因此在疫情爆发时期，二分类算法作为一种辅助诊断方法，具有很大的应用价值。Barstugan等人(2020)以机器学习算法对CT图像进行分类，该研究采用灰度共现矩阵、局部方向图、灰度游程矩阵、灰度区域大小矩阵和离散小波变换6种算法提取特征，用支持向量机(support vector machine，SVM)进行分类，在150幅CT图像上实验得出，最佳分类方法是以灰度区域大小矩阵进行特征提取，再用SVM分类，准确率为99.68 %。该研究体现了经典图像处理与机器学习算法仍有很大应用空间。

Chen等人(2020a)使用U-Net + +(Zhou等，2018)网络分割感染区域，再进行COVID-19阳性和阴性的分类，在46 096幅CT图像上实验，达到了98.85 %的准确率。Jin等人(2020)采用更复杂的3D UNet + +分割病变区域，并在分类任务中测试了ResNet-50(He等，2016)，Inception网络(Szegedy等，2016)，DPN-92(Chen等，2017)，Attention ResNet-50(Wang等，2017)，最终发现联合使用3D U-Net + +和ResNet-50可以达到最好的性能，在1 136个病例的胸部CT数据上实验，敏感性为97.4 %，特异性为92.2 %，AUC(area under curve)为0.991。综上可知采用2D网络分析CT切片图像，或用3D网络分析3维CT数据都可以达到很好的分类准确率。

结合CT数据的2D和3D特征可达到更好的效果。Gozes等人(2020)使用商用软件RADLogics Inc检测3D肺部CT扫描的结节和毛玻璃体混浊，再结合2D卷积神经网络进分割肺部区域并诊断是否患有COVID-19，在107个病例构成的数据集上，测试达到了0.996的AUC，98.2 %的敏感性，92.2 %的特异性。该算法敏感性高，更适合COVID-19病例的筛查。

3.1.2 三分类

COVID-19与非COVID-19肺炎相比，其临床症状和放射学特征都有部分重叠，因此，许多研究设计出三分类模型。Li等人(2020a)分割出了肺部区域作为预处理，再用COVnet(骨干网络为ResNet)网络进行分类，在4 356幅CT图像上实验，鉴定COVID-19的敏感性为90 %，特异性为96 %，AUC为0.96，这种设计的优点是不需要病变区域的标注信息。Zhang等人(2020a)采用DeepLabv3网络，首先分割出肺部区域，再分割出病变区域，最后使用调整后的ResNet-18网络在40 880个CT切片上实验，达到92.4 %的整体准确率。

由于COVID-19更容易引发胸膜附近的感染，且患者肺部通常不止一个独立的感染区域(Kanne，2020；Chung等，2020)。Butt等人(2020)根据这一特点，在分类网络中加入了位置注意力机制，用VNET-IR(inception-resnet)-RPN(region proposal network)17(Wu等，2019)分割网络和加入了位置注意力机制的ResNet-18分类网络，在618例CT样本上实验达到86.7 %的敏感性，81.3 %的精确率。

综上所述，采用分割网络如U-Net、V-Net及其变体分割病变区域，再结合分类网络如ResNet、VGG19等，并在大规模数据集上训练，均可得到较好的分类效果，在此基础上，可以结合生成对抗网络、位置注意力机制、迁移学习以及2D与3D特征结合等方法进一步提高准确率。

3.2 基于X光片的COVID-19诊断

Ai等人(2020)指出，在一些COVID-19爆发地区，可以将胸部X光片(chest X-Ray，CXR)检查作为COVID-19的主要筛查工具。Ng等人(2020)指出了感染COVID-19患者的X光片上的特征。此外，利用CXR成像进行COVID-19筛查有以下优势：实用性(CXR在大多数医疗保健系统中被视为标准设备)、便携性(便携式CXR系统可以在隔离室内执行成像，减少了成像过程中传播病毒的风险)。基于X光片的COVID-19诊断的工作有Wang等人(2020a)、Alqudah等人(2020b)、Zhang等人(2020)、Goodwin等人(2020)、Khobahi等人(2020)、Maghdid等人(2020a)、Hammoudi等人(2020)、Gaál等人(2020)、Narin等人(2020)和Hemdan等人(2020)。与基于CT的COVID-19诊断相同，面向X光片的算法根据分类结果也分为两部分，即二分类和三分类。

3.2.1 二分类

Hemdan等人(2020)采用VGG19、DenseNet201、InceptionV3、ResNetV2、InceptionResNetV2、Xception和MobileNetV2这7种经典的深度学习模型训练二分类网络，在50例X光片上实验发现，VGG19和DenseNet201达到了同样最好的性能，准确率都是83 %。Narin等人(2020)使用经过ImageNet数据集预训练的ResNet50模型，达到98 %的诊断准确率。实验表明使用迁移学习和更大的数据集，可以提高分类准确率。

3.2.2 三分类

由于COVID-19与其他肺炎的CXR有相似之处，而与正常人的胸片差别较大。Hammoudi等人(2020)设计了两阶段分类方法，第1阶段是用CNN模型来区分正常还是肺炎，在第2阶段，用CNN+RNN模型区分COVID-19和细菌性肺炎，准确率达到了90.7 %。Wang等人(2020a)采用Wong等人(2018)提出的生成合成器(generative synthesis)自动生成COVID-Net网络模型，测试准确率为83.5 %。

3.3 其他COVID-19诊断方法

由于COVID-19早期患者的放射学影像表现并不明显，根据临床信息(旅行和接触史、白细胞数、症状(发烧、咳嗽和痰液)，患者年龄和性别等)进行早期诊断是一种快速有效的方法。Langer等人(2020)收集了199个病例的74项临床信息，用人工神经网络算法达到了91.4 %的准确率，94.1 %的敏感性。临床医生在诊断的过程中，会综合考虑医学影像和临床信息两方面因素。根据这一特点，Mei等人(2020)使用CNN模型和多层感知机算法，结合CT影像和临床信息两类数据以诊断COVID-19，该算法敏感性为84.3 %，AUC为0.92，优于高级放射科医生(具有10年经验)的诊断结果。

值得注意的是，Maghdid等人(2020b)开发了基于智能手机的AI诊断系统，将CT影像和手机传感器数据(惯性传感器数据、咳嗽声和手指温度)上传到云服务器进行诊断。该系统具有便捷快速的优势，且随着该手机App的推广，数据集的规模也将不断增加，有助于进一步训练改进算法模型。

4 COVID-19的跟踪和预后

医学影像是医生跟踪观察病情的重要工具。语义分割网络可以分割出肺部感染区域并计算出感染率，作为医生预后的参考。Gozes等人(2020)根据CT切片的“特征图”和3维CT数据中检测到的不透明度测量值，提出“电晕评分”来衡量COVID-19患者的病情的变化。Hammoudi等人(2020)根据年龄，综合症和感染率来预测COVID-19患者的存活率。

除了根据医学图像预后，还有许多研究根据临床指标预测存活率。由于指标的数量多，变化复杂，需要智能算法来快速且敏锐地分析这些数据。Jiang等人(2020)根据丙氨酸转氨酶、肌痛和血红蛋白数，预测患者是否会发展为急性呼吸窘迫综合征(acute respiratory distress syndrome, ARDS)(COVID-19可能引发的一种致命并发症)，用最近邻算法和支持向量机算法达到80 %的准确率。Yan等人(2020)开发了基于XGBoost(eXtreme Gradient Boosting)算法的预后模型，从300多个特征中识别出3个关键的临床特征，预测COVID-19患者的存活率，准确率超过了90 %。这些预测算法，可以帮助医生及早发现高危患者并治疗，更好地分配医疗资源，降低死亡率，具有重大意义。

5 结语

算法专家和临床医生基于图像处理和机器学习等技术，提出各种计算机辅助诊断算法，来积极应对COVID-19大爆发带来的挑战。本文回顾了数月以来基于医学影像的AI诊断的研究成果。尽管有一些AI辅助诊断系统已经部署到医院中发挥实际作用，这些AI系统仍然具备很大的上升潜力，可以通过以下几个方面进一步提升其性能：

1) 人工智能作为数据驱动的一门学科，收集更多高质量的数据集，有助于进一步提升算法性能。

2) 现有的AI诊断技术多是基于医学影像的，诊断依据单一。结合临床信息、旅行接触史等多种类型数据可弥补仅用医学影像诊断的不足。

3) 由于非COVID-19肺炎和COVID-19有很多重叠的临床特征，如何更好地区分这两类肺炎也是未来的研究方向之一。

参考文献

  • Ai T, Yang Z L, Hou H Y, Zhan C N, Chen C, Lv W Z, Tao Q, Sun Z Y, Xia L M. 2020. Correlation of chest CT and RT-PCR testing for coronavirus disease 2019 (COVID-19) in China:a report of 1 014 cases. Radiology, 296(2): E32-E40 [DOI:10.1148/radiol.2020200642]
  • Allen Institute for AI, Chan Zuckerberg Initiative, Georgetown University's Center for Security and Emerging Technology, Microsoft Research, IBM, National Library of Medicine and The White House Office. 2020a. COVID-19 Open research dataset challenge (CORD-19)[DB/OL].[2020-05-01]. https://www.kaggle.com/allen-institute-for-ai/CORD-19-research-challenge
  • Allen Institute for AI, Chan Zuckerberg Initiative, Georgetown University's Center for Security and Emerging Technology, Microsoft Research, IBM, National Library of Medicine and The White House Office. 2020b. COVID-19 open research dataset[DB/OL].[2020-05-01]. https://www.semanticscholar.org/cord19
  • Alom M Z, Aspiras T, Taha T M and Asari V K. 2019. Skin cancer segmentation and classification with NABLA-N and inception recurrent residual convolutional networks[EB/OL].[2020-05-01]. https://arxiv.org/pdf/1904.11126.pdf
  • Alom M Z, Rahman M M S, Nasrin M S, Taha T M and Asari V K. 2020. COVID_MTNet: COVID-19 detection with multi-task deep learning approaches[EB/OL].[2020-05-01]. https://arxiv.org/pdf/2004.03747.pdf
  • Alqudah A M and Qazan S. 2020a. Augmented COVID-19 X-ray images dataset[DB/OL].[2020-05-01]. http://dx.doi.org/10.17632/2fxz4px6d8.4
  • Alqudah A M, Qazan S and Alqudah A. 2020b. Automated systems for detection of COVID-19 using chest X-ray images and lightweight convolutional neural networks[EB/OL].[2020-05-01]. https://europepmc.org/article/ppr/ppr155593
  • Bai H X, Wang R, Xiong Z, Hsieh B, Chang K, Halsey K, Tran T M L, Choi J W, Wang D C, Shi L B, Mei J, Jiang X L, Pan I, Zeng Q H, Hu P F, Li Y H, Fu F X, Huang R Y, Sebro R, Yu Q Z, Atalay M K and Liao W H. 2020. AI augmentation of radiologist performance in distinguishing COVID-19 from pneumonia of other etiology on chest CT. Radiology[EB/OL].[2020-05-01].https://pubs.rsna.org/doi/pdf/10.1148/radiol.2020201491[DOI:10.1148/radiol.2020201491]
  • Barstugan M, Ozkaya U and Ozturk S. 2020. Coronavirus (COVID-19) classification using CT images by machine learning methods[EB/OL].[2020-05-01]. https://arxiv.org/pdf/2003.09424.pdf
  • Born J, Brändle G, Cossio M, Disdier M, Goulet J, Roulin J and Wiedemann N. 2020. POCOVID-Net: automatic detection of COVID-19 from a new lung ultrasound imaging dataset (POCUS)[EB/OL].[2020-05-01]. https://arxiv.org/pdf/2004.12084.pdf
  • Butt C, Gill J, Chun D and Babu B A. 2020. Deep learning system to screen coronavirus disease 2019 pneumonia. Applied Intelligence.[EB/OL].[2020-05-01].https://doi.org/10.1007/s10489-020-01714-3[DOI:10.1007/s10489-020-01714-3]
  • Chen J, Wu L L, Zhang J, Zhang L, Gong D X, Zhao Y L, Hu S, Wang Y G, Hu X, Zheng B Q, Zhang K, Wu H L, Dong Z H, Xu Y M, Zhu Y J, Chen X, Yu L L and Yu H G. 2020a. Deep learning-based model for detecting 2019 novel coronavirus pneumonia on high-resolution computed tomography: a prospective study[EB/OL].[2020-05-01]. https://www.medrxiv.org/content/10.1101/2020.02.25.20021568v2
  • Chen N S, Zhou M, Dong X, Qu J M, Gong F Y, Han Y, Qiu Y, Wang J L, Liu Y, Wei Y, Xia J A, Yu T, Zhang X X, Zhang L. 2020b. Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China:a descriptive study. The Lancet, 395(10223): 507-513 [DOI:10.1016/S0140-6736(20)30211-7]
  • Chen Q Y, Allot A, Lu Z Y. 2020c. Keep up with the latest coronavirus research. Nature, 579(7798): #193 [DOI:10.1038/d41586-020-00694-1]
  • Chen Y P, Li J N, Xiao H X, Jin X J, Yan S C and Feng J S. 2017. Dual path networks//Proceedings of the 31st International Conference on Neural Information Processing Systems. Red Hook: Curran Associates Inc: 4470-4478
  • Chowdhury M E H, Rahman T, Khandakar A, Mazhar R, Kadir M A, Mahbub Z B, Islam K R, Khan M S, Iqbal A, Al-Emadi N, Reaz M B I, Islam T I. 2020. Can AI help in screening viral and COVID-19 pneumonia?. IEEE Access, 8: 132665-132676 [DOI:10.1109/ACCESS.2020.3010287]
  • Chung M, Bernheim A, Mei X Y, Zhang N, Huang M Q, Zeng X J, Cui J F, Xu W J, Yang Y, Fayad Z A, Jacobi A, Li K W, Li S L, Shan H. 2020. CT imaging features of 2019 novel coronavirus (2019-nCoV). Radiology, 295(1): 202-207 [DOI:10.1148/radiol.2020200230]
  • Cohen J P, Morrison P and Dao L. 2020. COVID-19 image data collection[EB/OL].[2020-05-01]. https://arxiv.org/pdf/2003.11597.pdf
  • Gaál G, Maga B and Lukács A. 2020. Attention U-Net based adversarial architectures for chest X-ray lung segmentation[EB/OL].[2020-05-01]. https://arxiv.org/pdf/2003.10304.pdf
  • Goodwin B D, Jaskolski C, Zhong C and Asmani H. 2020. Intra-model variability in COVID-19 classification using chest X-ray images[EB/OL].[2020-05-01]. https://arxiv.org/pdf/2005.02167.pdf
  • Gozes O, Frid-Adar M, Greenspan H, Browning P D, Zhang H Q, Ji W B, Bernheim A and Siegel E. 2020. Rapid AI development cycle for the coronavirus (COVID-19) pandemic: initial results for automated detection and patient monitoring using deep learning CT image analysis[EB/OL].[2020-05-01]. https://arxiv.org/pdf/2003.05037.pdf
  • Hammoudi K, Benhabiles H, Melkemi M, Dornaika F, Arganda-Carreras I, Collard D and Scherpereel A. 2020. Deep learning on chest X-ray images to detect and evaluate pneumonia cases at the era of COVID-19[EB/OL].[2020-05-01]. https://arxiv.org/pdf/2004.03399.pdf
  • He K M, Zhang X Y, Ren S Q and Sun J. 2016. Deep residual learning for image recognition//Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas: IEEE: 770-778[DOI:10.1109/CVPR.2016.90]
  • He X H, Yang X Y, Zhang S H, Zhao J Y, Zhang Y C, Xing E and Xie P T. 2020. Sample-efficient deep learning for COVID-19 diagnosis based on CT scans[EB/OL].[2020-05-01]. https://www.medrxiv.org/content/10.1101/2020.04.13.20063941v1
  • Hemdan E E D, Shouman M A and Karar M E. 2020. Covidx-Net: a framework of deep learning classifiers to diagnose COVID-19 in X-ray images[EB/OL].[2020-05-01]. https://arxiv.org/pdf/2003.11055.pdf
  • Hu S P, Gao Y, Niu Z M, Jiang Y H, Li L, Xiao X L, Wang M H, Fang E F, Menpes-Smith W, Xia J, Ye H and Yang G. 2020. Weakly supervised deep learning for COVID-19 infection detection and classification from CT images[EB/OL].[2020-05-01]. https://arxiv.org/pdf/2004.06689.pdf
  • Huang C L, Wang Y M, Li X W, Ren L L, Zhao J P, Hu Y, Zhang L, Fan G H, Xu J Y, Gu X Y, Cheng Z S, Yu T, Xia J A, Wei Y, Wu W J, Xie X L, Yin W, Li H, Liu M, Xiao Y, Gao H, Guo L, Xie J G, Wang G F, Jiang R M, Gao Z C, Jin Q, Wang J W, Cao B. 2020. Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China. The Lancet, 395(10223): 497-506 [DOI:10.1016/S0140-6736(20)30183-5]
  • Jiang X G, Coffee M, Bari A, Wang J Z, Jiang X Y, Huang J P, Shi J C, Dai J Y, Cai J, Zhang T X, Wu Z X, He G Q, Huang Y T. 2020. Towards an artificial intelligence framework for data-driven prediction of coronavirus clinical severity. Computers, Materials and Continua, 63(1): 537-551 [DOI:10.32604/cmc.2020.010691]
  • Jin S, Wang B, Xu H B, Luo C, Wei L, Zhao W, Hou X X, Ma W S, Xu Z Q, Zheng Z Z, Sun W B, Lan L, Zhang W, Mu X D, Shi C X, Wang Z X, Lee J, Jin Z J, Lin M G, Jin H B, Zhang L, Guo J, Zhao B Q, Ren Z Z, Wang S H, You Z, Dong J H, Wang X H, Wang J M and Xu W. 2020. AI-assisted CT imaging analysis for COVID-19 screening: building and deploying a medical AI system in four weeks[EB/OL].[2020-05-01]. https://www.medrxiv.org/content/10.1101/2020.03.19.20039354v1
  • Kanne J P. 2020. Chest CT findings in 2019 novel coronavirus (2019-nCoV) infections from Wuhan, China:key points for the radiologist. Radiology, 295(1): 16-17 [DOI:10.1148/radiol.2020200241]
  • Kermany D, Zhang K and Goldbaum M. 2018. Labeled optical coherence tomography (OCT) and chest X-ray images for classification[DB/OL].[2020-05-01]. https://data.mendeley.com/datasets/rscbjbr9sj/2
  • Khobahi S, Agarwal C and Soltanalian M. 2020. CoroNet: a deep network architecture for semi-supervised task-based identification of COVID-19 from chest X-Ray images[EB/OL].[2020-05-01]. https://www.medrxiv.org/content/10.1101/2020.04.14.20065722v1
  • Langer T, Favarato M, Giudici R, Bassi G, Garberi R, Villa F, Gay H, Zeduri A, Bragagnolo S, Molteni A, Beretta A, Corradin M, Moreno M, Vismara C, Perno C F, Buscema M, Grossi E and Fumagalli R. 2020. Use of machine learning to rapidly predict positivity to severe acute respiratory syndrome coronavirus 2 (SARS-COV-2) using basic clinical data[EB/OL].[2020-05-01]. https://www.researchsquare.com/article/rs-38576/v1
  • Li L, Qin L X, Xu Z G, Yin Y B, Wang X, Kong B, Bai J J, Lu Y, Fang Z H, Song Q, Cao K L, Liu D L, Wang G S, Xu Q Z, Fang X S, Zhang S Q, Xia J, Xia J. 2020a. Using artificial intelligence to detect COVID-19 and community-acquired pneumonia based on pulmonary CT:evaluation of the diagnostic accuracy. Radiology, 296(2): E65-E71 [DOI:10.1148/radiol.2020200905]
  • Li Z, Zhong Z, Li Y, Zhang T Y, Gao L X, Jin D K, Sun Y, Ye X H, Yu L, Hu Z Y, Xiao J, Huang L Y and Tang Y L. 2020b. From community acquired pneumonia to COVID-19: a deep learning based method for quantitative analysis of COVID-19 on thick-section CT scans[EB/OL].[2020-05-01]. https://www.medrxiv.org/content/10.1101/2020.04.17.20070219v1
  • Lu H Z, Stratton C W, Tang Y W. 2020. Outbreak of pneumonia of unknown etiology in Wuhan, China:the mystery and the miracle. Journal of Medical Virology, 92(4): 401-402 [DOI:10.1002/jmv.25678]
  • Maghdid H S, Asaad A T, Ghafoor K Z, Sadiq A S and Khan M K. 2020a. Diagnosing COVID-19 pneumonia from X-Ray and CT images using deep learning and transfer learning algorithms[EB/OL].[2020-05-01]. https://arxiv.org/pdf/2004.00038.pdf
  • Maghdid H S, Ghafoor K Z, Sadiq A S, Curran K, Rawat D B and Rabie K. 2020b. A novel AI-enabled framework to diagnose coronavirus COVID-19 using smartphone embedded sensors: design study[EB/OL].[2020-05-01]. https://arxiv.org/pdf/2003.07434.pdf
  • MedSeg. 2020. COVID-19 CT segmentation dataset[DB/OL].[2020-05-22]. http://medicalsegmentation.com/covid19/
  • Mei X Y, Lee H C, Diao K Y, Huang M Q, Lin B, Liu C Y, Xie Z Y, Ma Y X, Robson P M, Chung M, Bernheim A, Mani V, Calcagno C, Li K W, Li S L, Shan H, Lv J, Zhao T T, Xia J L, Long Q H, Steinberger S, Jacobi A, Deyer T, Luksza M, Liu F, Little B P, Fayad Z A and Yang Y. 2020. Artificial intelligence-enabled rapid diagnosis of patients with COVID-19. Nature Medicine.[EB/OL].[2020-05-01].https://doi.org/10.1038/s41591-020-0931-3[DOI:10.1038/s41591-020-0931-3]
  • Narin A, Kaya C and Pamuk Z. 2020. Automatic detection of coronavirus disease (COVID-19) using X-ray images and deep convolutional neural networks[EB/OL].[2020-05-18]. https://arxiv.org/pdf/2003.10849.pdf
  • National Health Commission of the People's Republic of China, National Administration of Traditional Chinese Medicine. 2020. New coronavirus pneumonia diagnosis and treatment plan (Trial Version 5 Revised)[EB/OL].[2020-05-18]. http://www.nhc.gov.cn/xcs/zhengcwj/202002/d4b895337e19445f8d728fcaf1e3e13a.shtml. (国家卫生健康委办公厅, 国家中医药管理局办公室. 2020.新型冠状病毒肺炎诊疗方案(试行第五版修正版)[EB/OL].[2020-05-01]. http://www.nhc.gov.cn/xcs/zhengcwj/202002/d4b895337e19445f8d728fcaf1e3e13a.shtml)
  • Ng M Y, Lee E Y, Yang J, Yang F F, Li X, Wang H X, Lui M M S, Lo S Y, Leung B, Khong P L, Hui C K M, Yuen K Y and Kuo M D. 2020. Imaging profile of the COVID-19 infection: radiologic findings and literature review. Radiology: Cardiothoracic Imaging[EB/OL].[2020-05-01].https://pubs.rsna.org/doi/10.1148/ryct.2020200034[DOI:10.1148/ryct.2020200034]
  • Shan F, Gao Y Z, Wang J, Shi W Y, Shi N N, Han M F, Xue Z, Shen D G and Shi Y X. 2020. Lung infection quantification of COVID-19 in CT images with deep learning[EB/OL].[2020-05-01]. https://arxiv.org/pdf/2003.04655.pdf
  • Shiraishi J, Katsuragawa S, Ikezoe J, Matsumoto T, Kobayashi T, Komatsu K I, Matsui M, Fujita H, Kodera Y, Doi K. 2000. Development of a digital image database for chest radiographs with and without a lung nodule:receiver operating characteristic analysis of radiologists' detection of pulmonary nodules. American Journal of Roentgenology, 174(1): 71-74 [DOI:10.2214/ajr.174.1.1740071]
  • Szegedy C, Vanhoucke V, Ioffe S, Shlens J and Wojna Z. 2016. Rethinking the inception architecture for computer vision//Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas: IEEE: 2818-2826[DOI:10.1109/CVPR.2016.308]
  • The British Society of Thoracic Imaging (BSTI). 2020. COVID-19 imaging database[DB/OL].[2020-05-22]. https://www.bsti.org.uk/training-and-education/covid-19-bsti-imaging-database/
  • Wang F, Jiang M Q, Qian C, Yang S, Li C, Zhang H G, Wang X G, Tang X O. 2017. Residual attention network for image classification//Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu: IEEE: 3156-3164[DOI:10.1109/CVPR.2017.683]
  • Wang L D and Wong A. 2020a. COVID-Net: a tailored deep convolutional neural network design for detection of COVID-19 cases from chest X-Ray images[EB/OL].[2020-05-13]. https://arxiv.org/pdf/2003.09871.pdf
  • Wang S, Kang B, Ma J L, Zeng X J, Xiao M M, Guo J, Cai M J, Yang J Y, Li Y D, Meng X F and Xu B. 2020. A deep learning algorithm using CT images to screen for Corona Virus Disease (COVID-19)[EB/OL].[2020-05-24]. https://www.medrxiv.org/content/10.1101/2020.02.14.20023028v5
  • Wong A, Shafiee M J, Chwyl B and Li F. 2020. Ferminets: learning generative machines to generate efficient neural networks via generative synthesis[EB/OL].[2020-05-14]. https://arxiv.org/pdf/1809.05989.pdf
  • World Health Organization (WHO). 2020a. WHO Director-General's remarks at the media briefing on 2019-nCoV on 11 February 2020[EB/OL].[2020-05-10]. https://www.who.int/dg/speeches/detail/who-director-general-s-remarks-at-the-media-briefing-on-2019-ncov-on-11-february-2020
  • World Health Organization (WHO). 2020b. Global research on coronavirus disease (COVID-19)[DB/OL].[2020-05-24]. https://www.who.int/emergencies/diseases/novel-coronavirus-2019/global-research-on-novel-coronavirus-2019-ncov
  • Wu W, Li X K, Du P, Lang G J, Xu M, Xu K J and Li L J. 2019. A deep learning system that generates quantitative CT reports for diagnosing pulmonary tuberculosis[EB/OL].[2020-05-13]. https://arxiv.org/pdf/1910.02285.pdf
  • Yan L, Zhang H T, Goncalves J, Xiao Y, Wang M L, Guo Y Q, Sun C, Tang X C, Jin L, Zhang M Y, Huang X, Xiao Y, Cao H S, Chen Y Y, Ren T X, Wang F, Xiao Y R, Huang S F, Tan X, Huang N N, Jiao B, Zhang Y, Luo A L, Mombaerts L, Jin J Y, Cao Z G, Li S S, Xu H and Yuan Y. 2020. A machine learning-based model for survival prediction in patients with severe COVID-19 infection[EB/OL].[2020-05-18]. https://www.medrxiv.org/content/10.1101/2020.02.27.20028027v3
  • Zhang K, Liu X H, Shen J, Li Z H, Sang Y, Wu X W, Zha Y F, Liang W H, Wang C D, Wang K, Ye L S, Gao M, Zhou Z G, Li L, Wang J, Yang Z H, Cai H M, Xu J, Yang L, Cai W J, Xu W Q, Wu S X, Zhang W, Jiang S P, Zheng L H, Zhang X, Wang L, Lu L, Li J M, Yin H P, Wang W, Li O L, Zhang C, Liang L, Wu T, Deng R Y, Wei K, Zhou Y, Chen T, Lau J Y N, Fok M, He J X, Lin T X, Li W M, Wang G Y. 2020a. Clinically applicable AI system for accurate diagnosis, quantitative measurements, and prognosis of COVID-19 pneumonia using computed tomography. Cell, 181(6): 1423-1433 [DOI:10.1016/j.cell.2020.04.045]
  • Zhang Y F, Niu S C, Qiu Z, Wei Y, Zhao P L, Yao J H, Huang J Z, Wu Q Y and Tan M K. 2020b. COVID-DA: deep domain adaptation from typical pneumonia to COVID-19[EB/OL].[2020-05-06]. https://arxiv.org/pdf/2005.01577.pdf
  • Zheng C S, Deng X B, Fu Q, Zhou Q, Feng J P, Ma H, Liu W Y and Wang X G. 2020. Deep learning-based detection for COVID-19 from chest CT using weak label[EB/OL].[2020-05-14]. https://www.medrxiv.org/content/10.1101/2020.03.12.20027185v2
  • Zhou Z W, Siddiquee M R, Tajbakhsh N and Liang J M. 2018. UNet++: a nested U-Net architecture for medical image segmentation//Proceedings of the 4th International Workshop on Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support. Granada, Spain: Springer: 3-11[DOI:10.1007/978-3-030-00889-5_1]