Iteratively Pruned Deep Learning Ensembles for COVID-19 Detection in Chest X-Rays

We demonstrate use of iteratively pruned deep learning model ensembles for detecting pulmonary manifestations of COVID-19 with chest X-rays. This disease is caused by the novel Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) virus, also known as the novel Coronavirus (2019-nCoV). A cust...

Descripción completa

Detalles Bibliográficos
Formato: Online Artículo Texto
Lenguaje:English
Publicado: IEEE 2020
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7394290/
https://www.ncbi.nlm.nih.gov/pubmed/32742893
http://dx.doi.org/10.1109/ACCESS.2020.3003810
_version_ 1783565204801454080
collection PubMed
description We demonstrate use of iteratively pruned deep learning model ensembles for detecting pulmonary manifestations of COVID-19 with chest X-rays. This disease is caused by the novel Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) virus, also known as the novel Coronavirus (2019-nCoV). A custom convolutional neural network and a selection of ImageNet pretrained models are trained and evaluated at patient-level on publicly available CXR collections to learn modality-specific feature representations. The learned knowledge is transferred and fine-tuned to improve performance and generalization in the related task of classifying CXRs as normal, showing bacterial pneumonia, or COVID-19-viral abnormalities. The best performing models are iteratively pruned to reduce complexity and improve memory efficiency. The predictions of the best-performing pruned models are combined through different ensemble strategies to improve classification performance. Empirical evaluations demonstrate that the weighted average of the best-performing pruned models significantly improves performance resulting in an accuracy of 99.01% and area under the curve of 0.9972 in detecting COVID-19 findings on CXRs. The combined use of modality-specific knowledge transfer, iterative model pruning, and ensemble learning resulted in improved predictions. We expect that this model can be quickly adopted for COVID-19 screening using chest radiographs.
format Online
Article
Text
id pubmed-7394290
institution National Center for Biotechnology Information
language English
publishDate 2020
publisher IEEE
record_format MEDLINE/PubMed
spelling pubmed-73942902020-07-31 Iteratively Pruned Deep Learning Ensembles for COVID-19 Detection in Chest X-Rays IEEE Access Biomedical Engineering We demonstrate use of iteratively pruned deep learning model ensembles for detecting pulmonary manifestations of COVID-19 with chest X-rays. This disease is caused by the novel Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) virus, also known as the novel Coronavirus (2019-nCoV). A custom convolutional neural network and a selection of ImageNet pretrained models are trained and evaluated at patient-level on publicly available CXR collections to learn modality-specific feature representations. The learned knowledge is transferred and fine-tuned to improve performance and generalization in the related task of classifying CXRs as normal, showing bacterial pneumonia, or COVID-19-viral abnormalities. The best performing models are iteratively pruned to reduce complexity and improve memory efficiency. The predictions of the best-performing pruned models are combined through different ensemble strategies to improve classification performance. Empirical evaluations demonstrate that the weighted average of the best-performing pruned models significantly improves performance resulting in an accuracy of 99.01% and area under the curve of 0.9972 in detecting COVID-19 findings on CXRs. The combined use of modality-specific knowledge transfer, iterative model pruning, and ensemble learning resulted in improved predictions. We expect that this model can be quickly adopted for COVID-19 screening using chest radiographs. IEEE 2020-06-19 /pmc/articles/PMC7394290/ /pubmed/32742893 http://dx.doi.org/10.1109/ACCESS.2020.3003810 Text en This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ https://creativecommons.org/licenses/by/4.0/This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
spellingShingle Biomedical Engineering
Iteratively Pruned Deep Learning Ensembles for COVID-19 Detection in Chest X-Rays
title Iteratively Pruned Deep Learning Ensembles for COVID-19 Detection in Chest X-Rays
title_full Iteratively Pruned Deep Learning Ensembles for COVID-19 Detection in Chest X-Rays
title_fullStr Iteratively Pruned Deep Learning Ensembles for COVID-19 Detection in Chest X-Rays
title_full_unstemmed Iteratively Pruned Deep Learning Ensembles for COVID-19 Detection in Chest X-Rays
title_short Iteratively Pruned Deep Learning Ensembles for COVID-19 Detection in Chest X-Rays
title_sort iteratively pruned deep learning ensembles for covid-19 detection in chest x-rays
topic Biomedical Engineering
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7394290/
https://www.ncbi.nlm.nih.gov/pubmed/32742893
http://dx.doi.org/10.1109/ACCESS.2020.3003810
work_keys_str_mv AT iterativelypruneddeeplearningensemblesforcovid19detectioninchestxrays
AT iterativelypruneddeeplearningensemblesforcovid19detectioninchestxrays
AT iterativelypruneddeeplearningensemblesforcovid19detectioninchestxrays
AT iterativelypruneddeeplearningensemblesforcovid19detectioninchestxrays
AT iterativelypruneddeeplearningensemblesforcovid19detectioninchestxrays
AT iterativelypruneddeeplearningensemblesforcovid19detectioninchestxrays