Implicit Neural Distance Representation for Unsupervised and Supervised Classification of Complex Anatomies

Kristine Aavild Juhl; Xabier Morales; Ole de Backer; Oscar Camara; Rasmus Reinhold Paulsen

doi:10.1007/978-3-030-87196-3_38

Implicit Neural Distance Representation for Unsupervised and Supervised Classification of Complex Anatomies

Kristine Aavild Juhl^*, Xabier Morales, Ole de Backer, Oscar Camara, Rasmus Reinhold Paulsen

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

Abstract

The task of 3D shape classification is closely related to finding a good representation of the shapes. In this study, we focus on surface representations of complex anatomies and on how such representations can be utilized for super- and unsupervised classification. We present a novel Implicit Neural Distance Representation based on unsigned distance fields (UDFs). The UDFs can be embedded into a low-dimensional latent space, which is optimized using only the shape itself. We demonstrate that this self-optimized latent space holds important global shape information useful for reconstructing the anatomies, but also that unsupervised clustering of the latent vectors successfully separates different anatomies (left atrium, left/right ear-canals and human faces). Finally, we show how the representation can be used to do gender classification of human face geometries, which is a notoriously hard problem.

Original language	English
Title of host publication	Medical Image Computing and Computer Assisted Intervention
Editors	Marleen de Bruijne, Philippe C. Cattin, Stéphane Cotin, Nicolas Padoy, Stefanie Speidel, Yefeng Zheng, Caroline Essert
Publisher	Springer
Publication date	2021
Pages	405-415
ISBN (Print)	9783030871956
DOIs	https://doi.org/10.1007/978-3-030-87196-3_38
Publication status	Published - 2021
Event	24^th International Conference on Medical Image Computing and Computer Assisted Intervention - Virtual, Online Duration: 27 Sept 2021 → 1 Oct 2021

Conference

Conference	24^th International Conference on Medical Image Computing and Computer Assisted Intervention
City	Virtual, Online
Period	27/09/2021 → 01/10/2021

Series	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	12902 LNCS
ISSN	0302-9743

Bibliographical note

Funding Information:
This work was supported by a PhD grant from the Technical University of Denmark-Department of Applied Mathematics and Computer Science (DTU Compute) and the Spanish Ministry of Science, Innovation and Universities under the Retos I+D Programme (RTI2018-101193-B-I00).

Publisher Copyright:
© 2021, Springer Nature Switzerland AG.

Keywords

Implicit functions
Shape analysis
Unsigned distance fields

Access to Document

10.1007/978-3-030-87196-3_38

OpenUrl availability

Full text

Cite this

Juhl, K. A., Morales, X., de Backer, O., Camara, O., & Paulsen, R. R. (2021). Implicit Neural Distance Representation for Unsupervised and Supervised Classification of Complex Anatomies. In M. de Bruijne, P. C. Cattin, S. Cotin, N. Padoy, S. Speidel, Y. Zheng, & C. Essert (Eds.), Medical Image Computing and Computer Assisted Intervention (pp. 405-415). Springer. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) Vol. 12902 LNCS https://doi.org/10.1007/978-3-030-87196-3_38

Juhl, Kristine Aavild ; Morales, Xabier ; de Backer, Ole et al. / Implicit Neural Distance Representation for Unsupervised and Supervised Classification of Complex Anatomies. Medical Image Computing and Computer Assisted Intervention. editor / Marleen de Bruijne ; Philippe C. Cattin ; Stéphane Cotin ; Nicolas Padoy ; Stefanie Speidel ; Yefeng Zheng ; Caroline Essert. Springer, 2021. pp. 405-415 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 12902 LNCS).

@inproceedings{6f5197db2ad44985954f6c9327632da6,

title = "Implicit Neural Distance Representation for Unsupervised and Supervised Classification of Complex Anatomies",

abstract = "The task of 3D shape classification is closely related to finding a good representation of the shapes. In this study, we focus on surface representations of complex anatomies and on how such representations can be utilized for super- and unsupervised classification. We present a novel Implicit Neural Distance Representation based on unsigned distance fields (UDFs). The UDFs can be embedded into a low-dimensional latent space, which is optimized using only the shape itself. We demonstrate that this self-optimized latent space holds important global shape information useful for reconstructing the anatomies, but also that unsupervised clustering of the latent vectors successfully separates different anatomies (left atrium, left/right ear-canals and human faces). Finally, we show how the representation can be used to do gender classification of human face geometries, which is a notoriously hard problem.",

keywords = "Implicit functions, Shape analysis, Unsigned distance fields",

author = "Juhl, {Kristine Aavild} and Xabier Morales and {de Backer}, Ole and Oscar Camara and Paulsen, {Rasmus Reinhold}",

note = "Funding Information: This work was supported by a PhD grant from the Technical University of Denmark-Department of Applied Mathematics and Computer Science (DTU Compute) and the Spanish Ministry of Science, Innovation and Universities under the Retos I+D Programme (RTI2018-101193-B-I00). Publisher Copyright: {\textcopyright} 2021, Springer Nature Switzerland AG.; 24<sup>th</sup> International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2021 ; Conference date: 27-09-2021 Through 01-10-2021",

year = "2021",

doi = "10.1007/978-3-030-87196-3_38",

language = "English",

isbn = "9783030871956",

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

publisher = "Springer",

pages = "405--415",

editor = "{de Bruijne}, Marleen and Cattin, {Philippe C.} and St{\'e}phane Cotin and Nicolas Padoy and Stefanie Speidel and Yefeng Zheng and Caroline Essert",

booktitle = "Medical Image Computing and Computer Assisted Intervention",

}

Juhl, KA, Morales, X, de Backer, O, Camara, O & Paulsen, RR 2021, Implicit Neural Distance Representation for Unsupervised and Supervised Classification of Complex Anatomies. in M de Bruijne, PC Cattin, S Cotin, N Padoy, S Speidel, Y Zheng & C Essert (eds), Medical Image Computing and Computer Assisted Intervention. Springer, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 12902 LNCS, pp. 405-415, 24^th International Conference on Medical Image Computing and Computer Assisted Intervention, Virtual, Online, 27/09/2021. https://doi.org/10.1007/978-3-030-87196-3_38

Implicit Neural Distance Representation for Unsupervised and Supervised Classification of Complex Anatomies. / Juhl, Kristine Aavild; Morales, Xabier; de Backer, Ole et al.
Medical Image Computing and Computer Assisted Intervention. ed. / Marleen de Bruijne; Philippe C. Cattin; Stéphane Cotin; Nicolas Padoy; Stefanie Speidel; Yefeng Zheng; Caroline Essert. Springer, 2021. p. 405-415 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 12902 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

TY - GEN

T1 - Implicit Neural Distance Representation for Unsupervised and Supervised Classification of Complex Anatomies

AU - Juhl, Kristine Aavild

AU - Morales, Xabier

AU - de Backer, Ole

AU - Camara, Oscar

AU - Paulsen, Rasmus Reinhold

N1 - Funding Information: This work was supported by a PhD grant from the Technical University of Denmark-Department of Applied Mathematics and Computer Science (DTU Compute) and the Spanish Ministry of Science, Innovation and Universities under the Retos I+D Programme (RTI2018-101193-B-I00). Publisher Copyright: © 2021, Springer Nature Switzerland AG.

PY - 2021

Y1 - 2021

N2 - The task of 3D shape classification is closely related to finding a good representation of the shapes. In this study, we focus on surface representations of complex anatomies and on how such representations can be utilized for super- and unsupervised classification. We present a novel Implicit Neural Distance Representation based on unsigned distance fields (UDFs). The UDFs can be embedded into a low-dimensional latent space, which is optimized using only the shape itself. We demonstrate that this self-optimized latent space holds important global shape information useful for reconstructing the anatomies, but also that unsupervised clustering of the latent vectors successfully separates different anatomies (left atrium, left/right ear-canals and human faces). Finally, we show how the representation can be used to do gender classification of human face geometries, which is a notoriously hard problem.

AB - The task of 3D shape classification is closely related to finding a good representation of the shapes. In this study, we focus on surface representations of complex anatomies and on how such representations can be utilized for super- and unsupervised classification. We present a novel Implicit Neural Distance Representation based on unsigned distance fields (UDFs). The UDFs can be embedded into a low-dimensional latent space, which is optimized using only the shape itself. We demonstrate that this self-optimized latent space holds important global shape information useful for reconstructing the anatomies, but also that unsupervised clustering of the latent vectors successfully separates different anatomies (left atrium, left/right ear-canals and human faces). Finally, we show how the representation can be used to do gender classification of human face geometries, which is a notoriously hard problem.

KW - Implicit functions

KW - Shape analysis

KW - Unsigned distance fields

U2 - 10.1007/978-3-030-87196-3_38

DO - 10.1007/978-3-030-87196-3_38

M3 - Article in proceedings

AN - SCOPUS:85116497787

SN - 9783030871956

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

SP - 405

EP - 415

BT - Medical Image Computing and Computer Assisted Intervention

A2 - de Bruijne, Marleen

A2 - Cattin, Philippe C.

A2 - Cotin, Stéphane

A2 - Padoy, Nicolas

A2 - Speidel, Stefanie

A2 - Zheng, Yefeng

A2 - Essert, Caroline

PB - Springer

T2 - 24<sup>th</sup> International Conference on Medical Image Computing and Computer Assisted Intervention

Y2 - 27 September 2021 through 1 October 2021

ER -

Juhl KA, Morales X, de Backer O, Camara O, Paulsen RR. Implicit Neural Distance Representation for Unsupervised and Supervised Classification of Complex Anatomies. In de Bruijne M, Cattin PC, Cotin S, Padoy N, Speidel S, Zheng Y, Essert C, editors, Medical Image Computing and Computer Assisted Intervention. Springer. 2021. p. 405-415. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 12902 LNCS). doi: 10.1007/978-3-030-87196-3_38

Implicit Neural Distance Representation for Unsupervised and Supervised Classification of Complex Anatomies

Abstract

Conference

Bibliographical note

Keywords

Access to Document

OpenUrl availability

Fingerprint

Cite this