AI for radiographic COVID-19 detection selects shortcuts over signal

Artificial intelligence (AI) researchers and radiologists have recently reported AI systems that accurately detect COVID-19 in chest radiographs. However, the robustness of these systems remains unclear. Using state-of-the-art techniques in explainable AI, we demonstrate that recent deep learning systems to detect COVID-19 from chest radiographs rely on confounding factors rather than medical pathology, creating an alarming situation in which the systems appear accurate, but fail when tested in new hospitals.

[1]  Loris Nanni,et al.  A critic evaluation of methods for COVID-19 automatic detection from X-ray images , 2020, Information Fusion.

[2]  Percy Liang,et al.  An Investigation of Why Overparameterization Exacerbates Spurious Correlations , 2020, ICML.

[3]  U. Rajendra Acharya,et al.  Automated detection of COVID-19 cases using deep neural networks with X-ray images , 2020, Computers in Biology and Medicine.

[4]  Pascal Sturmfels,et al.  Learning Explainable Models Using Attribution Priors , 2019, ArXiv.

[5]  Gustavo Carneiro,et al.  Detecting hip fractures with radiologist-level performance using deep neural networks , 2017, ArXiv.

[6]  Ciarán M Lee,et al.  Improving the accuracy of medical diagnosis with causal machine learning , 2020, Nature Communications.

[7]  Jonathan R. Medverd,et al.  Policies and Guidelines for COVID-19 Preparedness: Experiences from the University of Washington , 2020, Radiology.

[8]  Gabriel Erion,et al.  An adversarial approach for the robust classification of pneumonia from chest radiographs , 2020, CHIL.

[9]  Joseph Paul Cohen,et al.  COVID-19 Image Data Collection , 2020, ArXiv.

[10]  P. Sham,et al.  A note on the calculation of empirical P values from Monte Carlo procedures. , 2002, American journal of human genetics.

[11]  Andrea Laghi,et al.  Cautions about radiologic diagnosis of COVID-19 infection driven by artificial intelligence , 2020, The Lancet Digital Health.

[12]  Joseph Paul Cohen,et al.  COVID-19 Image Data Collection: Prospective Predictions Are the Future , 2020, The Journal of Machine Learning for Biomedical Imaging.

[13]  Antonio Pertusa,et al.  PadChest: A large chest x-ray image dataset with multi-label annotated reports , 2019, Medical Image Anal..

[14]  Dong-Hyun Kim,et al.  Multiple skin lesions diagnostics via integrated deep convolutional networks for segmentation and classification , 2020, Comput. Methods Programs Biomed..

[15]  Andrew Y. Ng,et al.  CheXNet: Radiologist-Level Pneumonia Detection on Chest X-Rays with Deep Learning , 2017, ArXiv.

[16]  Neil D. Lawrence,et al.  Dataset Shift in Machine Learning , 2009 .

[17]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[18]  Daniel C. Castro,et al.  Causality matters in medical imaging , 2019, Nature Communications.

[19]  N. Arun,et al.  Assessing the (Un)Trustworthiness of Saliency Maps for Localizing Abnormalities in Medical Imaging , 2020, medRxiv.

[20]  Ciarán M Lee,et al.  Publisher Correction: Improving the accuracy of medical diagnosis with causal machine learning , 2020, Nature communications.

[21]  Alexander Wong,et al.  COVID-Net: a tailored deep convolutional neural network design for detection of COVID-19 cases from chest X-ray images , 2020, Scientific reports.

[22]  Pascal Sturmfels,et al.  Visualizing the Impact of Feature Attribution Baselines , 2020 .

[23]  Marcus A. Badgeley,et al.  Variable generalization performance of a deep learning model to detect pneumonia in chest radiographs: A cross-sectional study , 2018, PLoS medicine.

[24]  Brian Pollack,et al.  Explanation by Progressive Exaggeration , 2020, ICLR.

[25]  François Laviolette,et al.  Domain-Adversarial Training of Neural Networks , 2015, J. Mach. Learn. Res..

[26]  B. Beck,et al.  Cross-sectional study , 2011 .

[27]  Ankur Taly,et al.  Axiomatic Attribution for Deep Networks , 2017, ICML.

[28]  Sameer Singh,et al.  Beyond Accuracy: Behavioral Testing of NLP Models with CheckList , 2020, ACL.

[29]  Allan Tucker,et al.  Estimating Uncertainty and Interpretability in Deep Learning for Coronavirus (COVID-19) Detection , 2020, ArXiv.

[30]  P. Lakhani,et al.  Deep Learning at Chest Radiography: Automated Classification of Pulmonary Tuberculosis by Using Convolutional Neural Networks. , 2017, Radiology.

[31]  Matthias Bethge,et al.  Shortcut Learning in Deep Neural Networks , 2020, Nat. Mach. Intell..

[32]  Charles E Kahn,et al.  How Might AI and Chest Imaging Help Unravel COVID-19’s Mysteries? , 2020, Radiology. Artificial intelligence.

[33]  M. Lenzen,et al.  Scientists’ warning on affluence , 2020, Nature Communications.

[34]  Antonella Santone,et al.  Explainable Deep Learning for Pulmonary Disease and Coronavirus COVID-19 Detection from X-rays , 2020, Computer Methods and Programs in Biomedicine.

[35]  Kilian Q. Weinberger,et al.  Densely Connected Convolutional Networks , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[36]  Subhashini Venugopalan,et al.  Detection of anaemia from retinal fundus images via deep learning , 2019, Nature Biomedical Engineering.

[37]  Mona G. Flores,et al.  Artificial intelligence for the detection of COVID-19 pneumonia on chest CT using multinational datasets , 2020, Nature Communications.

[38]  Jayashree Kalpathy-Cramer,et al.  Automated Assessment and Tracking of COVID-19 Pulmonary Disease Severity on Chest Radiographs using Convolutional Siamese Neural Networks , 2020, Radiology. Artificial intelligence.

[39]  Li Fei-Fei,et al.  ImageNet: A large-scale hierarchical image database , 2009, CVPR.

[40]  Pedagógia,et al.  Cross Sectional Study , 2019 .

[41]  K. Yuen,et al.  Imaging Profile of the COVID-19 Infection: Radiologic Findings and Literature Review , 2020, Radiology. Cardiothoracic imaging.

[42]  Ronald M. Summers,et al.  ChestX-ray: Hospital-Scale Chest X-ray Database and Benchmarks on Weakly Supervised Classification and Localization of Common Thorax Diseases , 2019, Deep Learning and Convolutional Neural Networks for Medical Imaging and Clinical Informatics.

[43]  Stefan Niehues,et al.  Comparing different deep learning architectures for classification of chest radiographs , 2020, Scientific Reports.

[44]  M. Kuo,et al.  Frequency and Distribution of Chest Radiographic Findings in COVID-19 Positive Patients , 2019, Radiology.

[45]  拓海 杉山,et al.  “Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks”の学習報告 , 2017 .

[46]  Gregory S. Corrado,et al.  Detecting Anemia from Retinal Fundus Images , 2019, ArXiv.

[47]  A. Ng,et al.  Deep-learning-assisted diagnosis for knee magnetic resonance imaging: Development and retrospective validation of MRNet , 2018, PLoS medicine.