Modeling Barrett’s Esophagus Progression Using Geometric Variational Autoencoders

Vivien van Veldhuizen; Sharvaree Vadgama; Onno de Boer; Sybren Meijer; Erik J. Bekkers

doi:https://doi.org/10.1007/978-3-031-45350-2_11

Modeling Barrett’s Esophagus Progression Using Geometric Variational Autoencoders

Vivien van Veldhuizen, Sharvaree Vadgama, Onno de Boer, Sybren Meijer, Erik J. Bekkers

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

Abstract

Early detection of Barrett’s Esophagus (BE), the only known precursor to Esophageal adenocarcinoma (EAC), is crucial for effectively preventing and treating esophageal cancer. In this work, we investigate the potential of geometric Variational Autoencoders (VAEs) to learn a meaningful latent representation that captures the progression of BE. We show that hyperspherical VAE (S -VAE ) and Kendall Shape VAE show improved classification accuracy, reconstruction loss, and generative capacity. Additionally, we present a novel autoencoder architecture that can generate qualitative images without the need for a variational framework while retaining the benefits of an autoencoder, such as improved stability and reconstruction quality.

Original language	English
Title of host publication	Cancer Prevention Through Early Detection - 2nd International Workshop, CaPTion 2023, Held in Conjunction with MICCAI 2023, Proceedings
Editors	Sharib Ali, Fons van der Sommen, Maureen van Eijnatten, Iris Kolenbrander, Bartłomiej W. Papież, Yueming Jin
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	132-142
Number of pages	11
Volume	14295 LNCS
ISBN (Print)	9783031453496
DOIs	https://doi.org/10.1007/978-3-031-45350-2_11
Publication status	Published - 2023
Event	2nd International Workshop on Cancer Prevention through early detecTion, CaPTion 2023 - Vancover, Canada Duration: 12 Oct 2023 → 12 Oct 2023

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	14295 LNCS

Conference

Conference	2nd International Workshop on Cancer Prevention through early detecTion, CaPTion 2023
Country/Territory	Canada
City	Vancover
Period	12/10/2023 → 12/10/2023

Keywords

Equivariance
Geometric Deep Learning
Oncology
Pathology
Representation Learning
Variational Autoencoders

Access to Document

https://doi.org/10.1007/978-3-031-45350-2_11

Cite this

van Veldhuizen, V., Vadgama, S., de Boer, O., Meijer, S., & Bekkers, E. J. (2023). Modeling Barrett’s Esophagus Progression Using Geometric Variational Autoencoders. In S. Ali, F. van der Sommen, M. van Eijnatten, I. Kolenbrander, B. W. Papież, & Y. Jin (Eds.), Cancer Prevention Through Early Detection - 2nd International Workshop, CaPTion 2023, Held in Conjunction with MICCAI 2023, Proceedings (Vol. 14295 LNCS, pp. 132-142). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14295 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-45350-2_11

van Veldhuizen, Vivien ; Vadgama, Sharvaree ; de Boer, Onno et al. / Modeling Barrett’s Esophagus Progression Using Geometric Variational Autoencoders. Cancer Prevention Through Early Detection - 2nd International Workshop, CaPTion 2023, Held in Conjunction with MICCAI 2023, Proceedings. editor / Sharib Ali ; Fons van der Sommen ; Maureen van Eijnatten ; Iris Kolenbrander ; Bartłomiej W. Papież ; Yueming Jin. Vol. 14295 LNCS Springer Science and Business Media Deutschland GmbH, 2023. pp. 132-142 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{c34100c5be7843dd82f2b24b2e50d458,

title = "Modeling Barrett{\textquoteright}s Esophagus Progression Using Geometric Variational Autoencoders",

abstract = "Early detection of Barrett{\textquoteright}s Esophagus (BE), the only known precursor to Esophageal adenocarcinoma (EAC), is crucial for effectively preventing and treating esophageal cancer. In this work, we investigate the potential of geometric Variational Autoencoders (VAEs) to learn a meaningful latent representation that captures the progression of BE. We show that hyperspherical VAE (S -VAE ) and Kendall Shape VAE show improved classification accuracy, reconstruction loss, and generative capacity. Additionally, we present a novel autoencoder architecture that can generate qualitative images without the need for a variational framework while retaining the benefits of an autoencoder, such as improved stability and reconstruction quality.",

keywords = "Equivariance, Geometric Deep Learning, Oncology, Pathology, Representation Learning, Variational Autoencoders",

author = "{van Veldhuizen}, Vivien and Sharvaree Vadgama and {de Boer}, Onno and Sybren Meijer and Bekkers, {Erik J.}",

note = "Publisher Copyright: {\textcopyright} 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.; 2nd International Workshop on Cancer Prevention through early detecTion, CaPTion 2023 ; Conference date: 12-10-2023 Through 12-10-2023",

year = "2023",

doi = "https://doi.org/10.1007/978-3-031-45350-2_11",

language = "English",

isbn = "9783031453496",

volume = "14295 LNCS",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "132--142",

editor = "Sharib Ali and {van der Sommen}, Fons and {van Eijnatten}, Maureen and Iris Kolenbrander and Papie{\.z}, {Bart{\l}omiej W.} and Yueming Jin",

booktitle = "Cancer Prevention Through Early Detection - 2nd International Workshop, CaPTion 2023, Held in Conjunction with MICCAI 2023, Proceedings",

address = "Germany",

}

van Veldhuizen, V, Vadgama, S, de Boer, O , Meijer, S & Bekkers, EJ 2023, Modeling Barrett’s Esophagus Progression Using Geometric Variational Autoencoders. in S Ali, F van der Sommen, M van Eijnatten, I Kolenbrander, BW Papież & Y Jin (eds), Cancer Prevention Through Early Detection - 2nd International Workshop, CaPTion 2023, Held in Conjunction with MICCAI 2023, Proceedings. vol. 14295 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 14295 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 132-142, 2nd International Workshop on Cancer Prevention through early detecTion, CaPTion 2023, Vancover, Canada, 12/10/2023. https://doi.org/10.1007/978-3-031-45350-2_11

Modeling Barrett’s Esophagus Progression Using Geometric Variational Autoencoders. / van Veldhuizen, Vivien; Vadgama, Sharvaree; de Boer, Onno et al.
Cancer Prevention Through Early Detection - 2nd International Workshop, CaPTion 2023, Held in Conjunction with MICCAI 2023, Proceedings. ed. / Sharib Ali; Fons van der Sommen; Maureen van Eijnatten; Iris Kolenbrander; Bartłomiej W. Papież; Yueming Jin. Vol. 14295 LNCS Springer Science and Business Media Deutschland GmbH, 2023. p. 132-142 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14295 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - Modeling Barrett’s Esophagus Progression Using Geometric Variational Autoencoders

AU - van Veldhuizen, Vivien

AU - Vadgama, Sharvaree

AU - de Boer, Onno

AU - Meijer, Sybren

AU - Bekkers, Erik J.

PY - 2023

Y1 - 2023

N2 - Early detection of Barrett’s Esophagus (BE), the only known precursor to Esophageal adenocarcinoma (EAC), is crucial for effectively preventing and treating esophageal cancer. In this work, we investigate the potential of geometric Variational Autoencoders (VAEs) to learn a meaningful latent representation that captures the progression of BE. We show that hyperspherical VAE (S -VAE ) and Kendall Shape VAE show improved classification accuracy, reconstruction loss, and generative capacity. Additionally, we present a novel autoencoder architecture that can generate qualitative images without the need for a variational framework while retaining the benefits of an autoencoder, such as improved stability and reconstruction quality.

AB - Early detection of Barrett’s Esophagus (BE), the only known precursor to Esophageal adenocarcinoma (EAC), is crucial for effectively preventing and treating esophageal cancer. In this work, we investigate the potential of geometric Variational Autoencoders (VAEs) to learn a meaningful latent representation that captures the progression of BE. We show that hyperspherical VAE (S -VAE ) and Kendall Shape VAE show improved classification accuracy, reconstruction loss, and generative capacity. Additionally, we present a novel autoencoder architecture that can generate qualitative images without the need for a variational framework while retaining the benefits of an autoencoder, such as improved stability and reconstruction quality.

KW - Equivariance

KW - Geometric Deep Learning

KW - Oncology

KW - Pathology

KW - Representation Learning

KW - Variational Autoencoders

UR - http://www.scopus.com/inward/record.url?scp=85176003265&partnerID=8YFLogxK

U2 - https://doi.org/10.1007/978-3-031-45350-2_11

DO - https://doi.org/10.1007/978-3-031-45350-2_11

M3 - Conference contribution

SN - 9783031453496

VL - 14295 LNCS

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 132

EP - 142

BT - Cancer Prevention Through Early Detection - 2nd International Workshop, CaPTion 2023, Held in Conjunction with MICCAI 2023, Proceedings

A2 - Ali, Sharib

A2 - van der Sommen, Fons

A2 - van Eijnatten, Maureen

A2 - Kolenbrander, Iris

A2 - Papież, Bartłomiej W.

A2 - Jin, Yueming

PB - Springer Science and Business Media Deutschland GmbH

T2 - 2nd International Workshop on Cancer Prevention through early detecTion, CaPTion 2023

Y2 - 12 October 2023 through 12 October 2023

ER -

van Veldhuizen V, Vadgama S, de Boer O , Meijer S, Bekkers EJ. Modeling Barrett’s Esophagus Progression Using Geometric Variational Autoencoders. In Ali S, van der Sommen F, van Eijnatten M, Kolenbrander I, Papież BW, Jin Y, editors, Cancer Prevention Through Early Detection - 2nd International Workshop, CaPTion 2023, Held in Conjunction with MICCAI 2023, Proceedings. Vol. 14295 LNCS. Springer Science and Business Media Deutschland GmbH. 2023. p. 132-142. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: https://doi.org/10.1007/978-3-031-45350-2_11

Modeling Barrett’s Esophagus Progression Using Geometric Variational Autoencoders

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Cite this