Accelerating Detection of Lung Pathologies with Explainable Ultrasound Image Analysis
Care during the COVID-19 pandemic hinges upon the existence of fast, safe, and highly sensitive diagnostic tools. Considering significant practical advantages of lung ultrasound (LUS) over other imaging techniques, but difficulties for doctors in pattern recognition, we aim to leverage machine learn...
Main Authors: | , , , , , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2021-01-01
|
Series: | Applied Sciences |
Subjects: | |
Online Access: | https://www.mdpi.com/2076-3417/11/2/672 |
_version_ | 1797412698564067328 |
---|---|
author | Jannis Born Nina Wiedemann Manuel Cossio Charlotte Buhre Gabriel Brändle Konstantin Leidermann Julie Goulet Avinash Aujayeb Michael Moor Bastian Rieck Karsten Borgwardt |
author_facet | Jannis Born Nina Wiedemann Manuel Cossio Charlotte Buhre Gabriel Brändle Konstantin Leidermann Julie Goulet Avinash Aujayeb Michael Moor Bastian Rieck Karsten Borgwardt |
author_sort | Jannis Born |
collection | DOAJ |
description | Care during the COVID-19 pandemic hinges upon the existence of fast, safe, and highly sensitive diagnostic tools. Considering significant practical advantages of lung ultrasound (LUS) over other imaging techniques, but difficulties for doctors in pattern recognition, we aim to leverage machine learning toward guiding diagnosis from LUS. We release the largest publicly available LUS dataset for COVID-19 consisting of 202 videos from four classes (COVID-19, bacterial pneumonia, non-COVID-19 viral pneumonia and healthy controls). On this dataset, we perform an in-depth study of the value of deep learning methods for the differential diagnosis of lung pathologies. We propose a frame-based model that correctly distinguishes COVID-19 LUS videos from healthy and bacterial pneumonia data with a sensitivity of <inline-formula><math display="inline"><semantics><mrow><mn>0.90</mn><mo>±</mo><mn>0.08</mn></mrow></semantics></math></inline-formula> and a specificity of <inline-formula><math display="inline"><semantics><mrow><mn>0.96</mn><mo>±</mo><mn>0.04</mn></mrow></semantics></math></inline-formula>. To investigate the utility of the proposed method, we employ interpretability methods for the spatio-temporal localization of pulmonary biomarkers, which are deemed useful for human-in-the-loop scenarios in a blinded study with medical experts. Aiming for robustness, we perform uncertainty estimation and demonstrate the model to recognize low-confidence situations which also improves performance. Lastly, we validated our model on an independent test dataset and report promising performance (sensitivity <inline-formula><math display="inline"><semantics><mrow><mn>0.806</mn></mrow></semantics></math></inline-formula>, specificity <inline-formula><math display="inline"><semantics><mrow><mn>0.962</mn></mrow></semantics></math></inline-formula>). The provided dataset facilitates the validation of related methodology in the community and the proposed framework might aid the development of a fast, accessible screening method for pulmonary diseases. Dataset and all code are publicly available at: https://github.com/BorgwardtLab/covid19_ultrasound. |
first_indexed | 2024-03-09T05:06:57Z |
format | Article |
id | doaj.art-6093c3b4857349e59a5b164bba9189fe |
institution | Directory Open Access Journal |
issn | 2076-3417 |
language | English |
last_indexed | 2024-03-09T05:06:57Z |
publishDate | 2021-01-01 |
publisher | MDPI AG |
record_format | Article |
series | Applied Sciences |
spelling | doaj.art-6093c3b4857349e59a5b164bba9189fe2023-12-03T12:53:53ZengMDPI AGApplied Sciences2076-34172021-01-0111267210.3390/app11020672Accelerating Detection of Lung Pathologies with Explainable Ultrasound Image AnalysisJannis Born0Nina Wiedemann1Manuel Cossio2Charlotte Buhre3Gabriel Brändle4Konstantin Leidermann5Julie Goulet6Avinash Aujayeb7Michael Moor8Bastian Rieck9Karsten Borgwardt10Department of Biosystems Science and Engineering, ETH Zurich, 4058 Basel, SwitzerlandDepartment of Computer Science, ETH Zurich, 8092 Zurich, SwitzerlandDepartment of Mathematics and Computer Science, University of Barcelona, 08007 Barcelona, SpainBrandenburg Medical School Theodor Fontane, 16816 Neuruppin, GermanyPediatric Emergency Department, Hirslanden Clinique des Grangettes, 1224 Geneva, SwitzerlandDepartment of Philosophy, University of Vienna, 1010 Vienna, AustriaPhysik Department T35 and Bernstein Center for Computational Neuroscience, Technische Universität München, 85747 Garching bei München, GermanyNorthumbria Specialist Emergency Care Hospital, Cramlington NE23 6NZ, UKDepartment of Biosystems Science and Engineering, ETH Zurich, 4058 Basel, SwitzerlandDepartment of Biosystems Science and Engineering, ETH Zurich, 4058 Basel, SwitzerlandDepartment of Biosystems Science and Engineering, ETH Zurich, 4058 Basel, SwitzerlandCare during the COVID-19 pandemic hinges upon the existence of fast, safe, and highly sensitive diagnostic tools. Considering significant practical advantages of lung ultrasound (LUS) over other imaging techniques, but difficulties for doctors in pattern recognition, we aim to leverage machine learning toward guiding diagnosis from LUS. We release the largest publicly available LUS dataset for COVID-19 consisting of 202 videos from four classes (COVID-19, bacterial pneumonia, non-COVID-19 viral pneumonia and healthy controls). On this dataset, we perform an in-depth study of the value of deep learning methods for the differential diagnosis of lung pathologies. We propose a frame-based model that correctly distinguishes COVID-19 LUS videos from healthy and bacterial pneumonia data with a sensitivity of <inline-formula><math display="inline"><semantics><mrow><mn>0.90</mn><mo>±</mo><mn>0.08</mn></mrow></semantics></math></inline-formula> and a specificity of <inline-formula><math display="inline"><semantics><mrow><mn>0.96</mn><mo>±</mo><mn>0.04</mn></mrow></semantics></math></inline-formula>. To investigate the utility of the proposed method, we employ interpretability methods for the spatio-temporal localization of pulmonary biomarkers, which are deemed useful for human-in-the-loop scenarios in a blinded study with medical experts. Aiming for robustness, we perform uncertainty estimation and demonstrate the model to recognize low-confidence situations which also improves performance. Lastly, we validated our model on an independent test dataset and report promising performance (sensitivity <inline-formula><math display="inline"><semantics><mrow><mn>0.806</mn></mrow></semantics></math></inline-formula>, specificity <inline-formula><math display="inline"><semantics><mrow><mn>0.962</mn></mrow></semantics></math></inline-formula>). The provided dataset facilitates the validation of related methodology in the community and the proposed framework might aid the development of a fast, accessible screening method for pulmonary diseases. Dataset and all code are publicly available at: https://github.com/BorgwardtLab/covid19_ultrasound.https://www.mdpi.com/2076-3417/11/2/672computer visionConvolutional neural networkCOVID-19deep learninginterpretabilitypneumonia |
spellingShingle | Jannis Born Nina Wiedemann Manuel Cossio Charlotte Buhre Gabriel Brändle Konstantin Leidermann Julie Goulet Avinash Aujayeb Michael Moor Bastian Rieck Karsten Borgwardt Accelerating Detection of Lung Pathologies with Explainable Ultrasound Image Analysis Applied Sciences computer vision Convolutional neural network COVID-19 deep learning interpretability pneumonia |
title | Accelerating Detection of Lung Pathologies with Explainable Ultrasound Image Analysis |
title_full | Accelerating Detection of Lung Pathologies with Explainable Ultrasound Image Analysis |
title_fullStr | Accelerating Detection of Lung Pathologies with Explainable Ultrasound Image Analysis |
title_full_unstemmed | Accelerating Detection of Lung Pathologies with Explainable Ultrasound Image Analysis |
title_short | Accelerating Detection of Lung Pathologies with Explainable Ultrasound Image Analysis |
title_sort | accelerating detection of lung pathologies with explainable ultrasound image analysis |
topic | computer vision Convolutional neural network COVID-19 deep learning interpretability pneumonia |
url | https://www.mdpi.com/2076-3417/11/2/672 |
work_keys_str_mv | AT jannisborn acceleratingdetectionoflungpathologieswithexplainableultrasoundimageanalysis AT ninawiedemann acceleratingdetectionoflungpathologieswithexplainableultrasoundimageanalysis AT manuelcossio acceleratingdetectionoflungpathologieswithexplainableultrasoundimageanalysis AT charlottebuhre acceleratingdetectionoflungpathologieswithexplainableultrasoundimageanalysis AT gabrielbrandle acceleratingdetectionoflungpathologieswithexplainableultrasoundimageanalysis AT konstantinleidermann acceleratingdetectionoflungpathologieswithexplainableultrasoundimageanalysis AT juliegoulet acceleratingdetectionoflungpathologieswithexplainableultrasoundimageanalysis AT avinashaujayeb acceleratingdetectionoflungpathologieswithexplainableultrasoundimageanalysis AT michaelmoor acceleratingdetectionoflungpathologieswithexplainableultrasoundimageanalysis AT bastianrieck acceleratingdetectionoflungpathologieswithexplainableultrasoundimageanalysis AT karstenborgwardt acceleratingdetectionoflungpathologieswithexplainableultrasoundimageanalysis |