Explainable-by-design Semi-Supervised Representation Learning for COVID-19 Diagnosis from CT Imaging

Abel Díaz Berenguer; Hichem Sahli; Boris Joukovsky; Maryna Kvasnytsia; Ine Dirks; Mitchel Alioscha-Pérez; Nikos Deligiannis; Panagiotis Gonidakis; Sebastián Amador Sánchez; Redona Brahimetaj; Evgenia Papavasileiou; Jonathan Cheung-Wai Chan; Fei Li; Shangzhen Song; Yixin Yang; Sofie Tilborghs; Siri Willems; Tom Eelbode; Jeroen Bertels; Dirk Vandermeulen; Frederik Maes; Paul Suetens; Lucas Fidon; Tom Vercauteren; David Robben; Arne Brys; Dirk Smeets; Bart Ilsen; Nico Buls; Nina Watté; Johan de Mey; Annemiek Snoeckx; Paul M. Parizel; Julien Guiot; Louis Deprez; Paul Meunier; Stefaan Gryspeerdt; Kristof De Smet; Bart Jansen; Jef Vandemeulebroucke

Explainable-by-design Semi-Supervised Representation Learning for COVID-19 Diagnosis from CT Imaging

Abel Díaz Berenguer
, Hichem Sahli
, Boris Joukovsky
, Maryna Kvasnytsia
, Ine Dirks
, Mitchel Alioscha-Pérez
, Nikos Deligiannis
, Panagiotis Gonidakis
, Sebastián Amador Sánchez
, Redona Brahimetaj
, Evgenia Papavasileiou
, Jonathan Cheung-Wai Chan
, Fei Li
, Shangzhen Song
, Yixin Yang
, Sofie Tilborghs
, Siri Willems
, Tom Eelbode
, Jeroen Bertels
, Dirk Vandermeulen

Frederik Maes, Paul Suetens, Lucas Fidon, Tom Vercauteren, David Robben, Arne Brys, Dirk Smeets, Bart Ilsen, Nico Buls, Nina Watté, Johan de Mey, Annemiek Snoeckx, Paul M. Parizel, Julien Guiot, Louis Deprez, Paul Meunier, Stefaan Gryspeerdt, Kristof De Smet, Bart Jansen, Jef Vandemeulebroucke

Research output: Working paper › Preprint

Abstract

Our motivating application is a real-world problem: COVID-19 classification from CT imaging, for which we present a explainable Deep Learning approach based on a semi-supervised classification pipeline that employs variational
autoencoders to extract efficient feature embedding. We have optimized the architecture of two different networks for CT images: (i) a novel conditional variational autoencoder (CVAE) with a specific architecture that integrates the
class labels inside the encoder layers and uses side information with shared attention layers for the encoder, which make the most of the contextual clues for representation learning, and (ii) a downstream convolutional neural network
for supervised classification using the encoder structure of the CVAE. With the explainable classification results, the proposed diagnosis system is very effective for COVID-19 classification. Based on the promising results obtained
qualitatively and quantitatively, we envisage a wide deployment of our developed technique in large-scale clinical studies. Code is available at https://git.etrovub.be/AVSP/ct-based-covid-19-diagnostic-tool.git

Original language	English
Publisher	arXiv
Volume	abs/2011.11719
Publication status	Published - 2 Sept 2021

Publication series

Name	CoRR

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 3 Good Health and Well-being

Access to Document

https://arxiv.org/abs/2011.11719Licence: CC BY-NC-ND

Cite this

Berenguer, A. D., Sahli, H., Joukovsky, B., Kvasnytsia, M., Dirks, I., Alioscha-Pérez, M., Deligiannis, N., Gonidakis, P., Sánchez, S. A., Brahimetaj, R., Papavasileiou, E., Chan, J. C.-W., Li, F., Song, S., Yang, Y., Tilborghs, S., Willems, S., Eelbode, T., Bertels, J., ... Vandemeulebroucke, J. (2021). Explainable-by-design Semi-Supervised Representation Learning for COVID-19 Diagnosis from CT Imaging. (CoRR). arXiv. https://arxiv.org/abs/2011.11719

@techreport{114bd2ba820342b48d098500b9b96912,

title = "Explainable-by-design Semi-Supervised Representation Learning for COVID-19 Diagnosis from CT Imaging",

abstract = "Our motivating application is a real-world problem: COVID-19 classification from CT imaging, for which we present a explainable Deep Learning approach based on a semi-supervised classification pipeline that employs variational autoencoders to extract efficient feature embedding. We have optimized the architecture of two different networks for CT images: (i) a novel conditional variational autoencoder (CVAE) with a specific architecture that integrates the class labels inside the encoder layers and uses side information with shared attention layers for the encoder, which make the most of the contextual clues for representation learning, and (ii) a downstream convolutional neural network for supervised classification using the encoder structure of the CVAE. With the explainable classification results, the proposed diagnosis system is very effective for COVID-19 classification. Based on the promising results obtained qualitatively and quantitatively, we envisage a wide deployment of our developed technique in large-scale clinical studies. Code is available at https://git.etrovub.be/AVSP/ct-based-covid-19-diagnostic-tool.git",

author = "Berenguer, \{Abel D{\'i}az\} and Hichem Sahli and Boris Joukovsky and Maryna Kvasnytsia and Ine Dirks and Mitchel Alioscha-P{\'e}rez and Nikos Deligiannis and Panagiotis Gonidakis and S{\'a}nchez, \{Sebasti{\'a}n Amador\} and Redona Brahimetaj and Evgenia Papavasileiou and Chan, \{Jonathan Cheung-Wai\} and Fei Li and Shangzhen Song and Yixin Yang and Sofie Tilborghs and Siri Willems and Tom Eelbode and Jeroen Bertels and Dirk Vandermeulen and Frederik Maes and Paul Suetens and Lucas Fidon and Tom Vercauteren and David Robben and Arne Brys and Dirk Smeets and Bart Ilsen and Nico Buls and Nina Watt{\'e} and Mey, \{Johan de\} and Annemiek Snoeckx and Parizel, \{Paul M.\} and Julien Guiot and Louis Deprez and Paul Meunier and Stefaan Gryspeerdt and Smet, \{Kristof De\} and Bart Jansen and Jef Vandemeulebroucke",

year = "2021",

month = sep,

day = "2",

language = "English",

volume = "abs/2011.11719",

series = "CoRR",

publisher = "arXiv",

address = "United States",

type = "WorkingPaper",

institution = "arXiv",

}

Berenguer, AD, Sahli, H, Joukovsky, B, Kvasnytsia, M, Dirks, I, Alioscha-Pérez, M, Deligiannis, N, Gonidakis, P, Sánchez, SA, Brahimetaj, R, Papavasileiou, E, Chan, JC-W, Li, F, Song, S, Yang, Y, Tilborghs, S, Willems, S, Eelbode, T, Bertels, J, Vandermeulen, D, Maes, F, Suetens, P, Fidon, L, Vercauteren, T, Robben, D, Brys, A, Smeets, D, Ilsen, B, Buls, N, Watté, N, Mey, JD, Snoeckx, A, Parizel, PM, Guiot, J, Deprez, L, Meunier, P, Gryspeerdt, S, Smet, KD, Jansen, B & Vandemeulebroucke, J 2021 'Explainable-by-design Semi-Supervised Representation Learning for COVID-19 Diagnosis from CT Imaging' CoRR, arXiv. <https://arxiv.org/abs/2011.11719>

TY - UNPB

T1 - Explainable-by-design Semi-Supervised Representation Learning for COVID-19 Diagnosis from CT Imaging

AU - Berenguer, Abel Díaz

AU - Sahli, Hichem

AU - Joukovsky, Boris

AU - Kvasnytsia, Maryna

AU - Dirks, Ine

AU - Alioscha-Pérez, Mitchel

AU - Deligiannis, Nikos

AU - Gonidakis, Panagiotis

AU - Sánchez, Sebastián Amador

AU - Brahimetaj, Redona

AU - Papavasileiou, Evgenia

AU - Chan, Jonathan Cheung-Wai

AU - Li, Fei

AU - Song, Shangzhen

AU - Yang, Yixin

AU - Tilborghs, Sofie

AU - Willems, Siri

AU - Eelbode, Tom

AU - Bertels, Jeroen

AU - Vandermeulen, Dirk

AU - Maes, Frederik

AU - Suetens, Paul

AU - Fidon, Lucas

AU - Vercauteren, Tom

AU - Robben, David

AU - Brys, Arne

AU - Smeets, Dirk

AU - Ilsen, Bart

AU - Buls, Nico

AU - Watté, Nina

AU - Mey, Johan de

AU - Snoeckx, Annemiek

AU - Parizel, Paul M.

AU - Guiot, Julien

AU - Deprez, Louis

AU - Meunier, Paul

AU - Gryspeerdt, Stefaan

AU - Smet, Kristof De

AU - Jansen, Bart

AU - Vandemeulebroucke, Jef

PY - 2021/9/2

Y1 - 2021/9/2

N2 - Our motivating application is a real-world problem: COVID-19 classification from CT imaging, for which we present a explainable Deep Learning approach based on a semi-supervised classification pipeline that employs variational autoencoders to extract efficient feature embedding. We have optimized the architecture of two different networks for CT images: (i) a novel conditional variational autoencoder (CVAE) with a specific architecture that integrates the class labels inside the encoder layers and uses side information with shared attention layers for the encoder, which make the most of the contextual clues for representation learning, and (ii) a downstream convolutional neural network for supervised classification using the encoder structure of the CVAE. With the explainable classification results, the proposed diagnosis system is very effective for COVID-19 classification. Based on the promising results obtained qualitatively and quantitatively, we envisage a wide deployment of our developed technique in large-scale clinical studies. Code is available at https://git.etrovub.be/AVSP/ct-based-covid-19-diagnostic-tool.git

AB - Our motivating application is a real-world problem: COVID-19 classification from CT imaging, for which we present a explainable Deep Learning approach based on a semi-supervised classification pipeline that employs variational autoencoders to extract efficient feature embedding. We have optimized the architecture of two different networks for CT images: (i) a novel conditional variational autoencoder (CVAE) with a specific architecture that integrates the class labels inside the encoder layers and uses side information with shared attention layers for the encoder, which make the most of the contextual clues for representation learning, and (ii) a downstream convolutional neural network for supervised classification using the encoder structure of the CVAE. With the explainable classification results, the proposed diagnosis system is very effective for COVID-19 classification. Based on the promising results obtained qualitatively and quantitatively, we envisage a wide deployment of our developed technique in large-scale clinical studies. Code is available at https://git.etrovub.be/AVSP/ct-based-covid-19-diagnostic-tool.git

UR - https://dblp.org/db/journals/corr/corr2011.html#abs-2011-11719

M3 - Preprint

VL - abs/2011.11719

T3 - CoRR

BT - Explainable-by-design Semi-Supervised Representation Learning for COVID-19 Diagnosis from CT Imaging

PB - arXiv

ER -

Explainable-by-design Semi-Supervised Representation Learning for COVID-19 Diagnosis from CT Imaging

Abstract

Publication series

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this