Geometric Deep Learning for the Assessment of Thrombosis Risk in the Left Atrial Appendage

Xabier Morales; Jordi Mill; Guillem Simeon; Kristine A. Juhl; Ole De Backer; Rasmus R. Paulsen; Oscar Camara

doi:10.1007/978-3-030-78710-3_61

Geometric Deep Learning for the Assessment of Thrombosis Risk in the Left Atrial Appendage

Xabier Morales^*, Jordi Mill, Guillem Simeon, Kristine A. Juhl, Ole De Backer, Rasmus R. Paulsen, Oscar Camara

^*Corresponding author for this work

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

46 Downloads (Pure)

Abstract

The assessment of left atrial appendage (LAA) thrombogenesis has experienced major advances with the adoption of patient-specific computational fluid dynamics (CFD) simulations. Nonetheless, due to the vast computational resources and long execution times required by fluid dynamics solvers, there is an ever-growing body of work aiming to develop surrogate models of fluid flow simulations based on neural networks. The present study builds on this foundation by developing a deep learning (DL) framework capable of predicting the endothelial cell activation potential (ECAP), linked to the risk of thrombosis, solely from the patient-specific LAA geometry. To this end, we leveraged recent advancements in Geometric DL, which seamlessly extend the unparalleled potential of convolutional neural networks (CNN), to non-Euclidean data such as meshes. The model was trained with a dataset combining 202 synthetic and 54 real LAA, predicting the ECAP distributions instantaneously, with an average mean absolute error of 0.563. Moreover, the resulting framework manages to predict the anatomical features related to higher ECAP values even when trained exclusively on synthetic cases.

Original language	English
Title of host publication	Functional Imaging and Modeling of the Heart
Editors	Daniel B. Ennis, Luigi E. Perotti, Vicky Y. Wang
Publisher	Springer
Publication date	2021
Pages	639-649
ISBN (Print)	9783030787097
DOIs	https://doi.org/10.1007/978-3-030-78710-3_61
Publication status	Published - 2021
Event	11^th International Conference on Functional Imaging and Modeling of the Heart - Virtual, Online Duration: 21 Jun 2021 → 25 Jun 2021 https://fimh2021.github.io/

Conference

Conference	11^th International Conference on Functional Imaging and Modeling of the Heart
City	Virtual, Online
Period	21/06/2021 → 25/06/2021
Internet address	https://fimh2021.github.io/

Series	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	12738 LNCS
ISSN	0302-9743

Bibliographical note

Funding Information:
Funding. This work was supported by the Agency for Management of University and Research Grants of the Generalitat de Catalunya under the Grants for the Contracting of New Research Staff Programme - FI (2020-FI-B-00690) and the Spanish Ministry of Economy and Competitiveness under the Programme for the Formation of Doctors (PRE2018-084062), the Maria de Maeztu Units of Excellence Programme (MDM-2015-0502) and the Retos Investigación project (RTI2018-101193-B-I00). Additionally, this work was supported by the H2020 EU SimCardioTest project (Digital transformation in Health and Care SC1-DTH-06-2020; grant agreement No. 101016496).

Publisher Copyright:
© 2021, Springer Nature Switzerland AG.

Keywords

Computational fluid dynamics
Geometric deep learning
Left atrial appendage
Thrombus formation

Access to Document

10.1007/978-3-030-78710-3_61

FulltextAccepted author manuscript, 4.09 MB

OpenUrl availability

Full text

Cite this

Morales, Xabier ; Mill, Jordi ; Simeon, Guillem et al. / Geometric Deep Learning for the Assessment of Thrombosis Risk in the Left Atrial Appendage. Functional Imaging and Modeling of the Heart. editor / Daniel B. Ennis ; Luigi E. Perotti ; Vicky Y. Wang. Springer, 2021. pp. 639-649 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 12738 LNCS).

@inproceedings{63e386102f7749239338c325ae237b85,

title = "Geometric Deep Learning for the Assessment of Thrombosis Risk in the Left Atrial Appendage",

abstract = "The assessment of left atrial appendage (LAA) thrombogenesis has experienced major advances with the adoption of patient-specific computational fluid dynamics (CFD) simulations. Nonetheless, due to the vast computational resources and long execution times required by fluid dynamics solvers, there is an ever-growing body of work aiming to develop surrogate models of fluid flow simulations based on neural networks. The present study builds on this foundation by developing a deep learning (DL) framework capable of predicting the endothelial cell activation potential (ECAP), linked to the risk of thrombosis, solely from the patient-specific LAA geometry. To this end, we leveraged recent advancements in Geometric DL, which seamlessly extend the unparalleled potential of convolutional neural networks (CNN), to non-Euclidean data such as meshes. The model was trained with a dataset combining 202 synthetic and 54 real LAA, predicting the ECAP distributions instantaneously, with an average mean absolute error of 0.563. Moreover, the resulting framework manages to predict the anatomical features related to higher ECAP values even when trained exclusively on synthetic cases.",

keywords = "Computational fluid dynamics, Geometric deep learning, Left atrial appendage, Thrombus formation",

author = "Xabier Morales and Jordi Mill and Guillem Simeon and Juhl, {Kristine A.} and {De Backer}, Ole and Paulsen, {Rasmus R.} and Oscar Camara",

note = "Funding Information: Funding. This work was supported by the Agency for Management of University and Research Grants of the Generalitat de Catalunya under the Grants for the Contracting of New Research Staff Programme - FI (2020-FI-B-00690) and the Spanish Ministry of Economy and Competitiveness under the Programme for the Formation of Doctors (PRE2018-084062), the Maria de Maeztu Units of Excellence Programme (MDM-2015-0502) and the Retos Investigaci{\'o}n project (RTI2018-101193-B-I00). Additionally, this work was supported by the H2020 EU SimCardioTest project (Digital transformation in Health and Care SC1-DTH-06-2020; grant agreement No. 101016496). Publisher Copyright: {\textcopyright} 2021, Springer Nature Switzerland AG.; 11<sup>th</sup> International Conference on Functional Imaging and Modeling of the Heart, FIMH 2021 ; Conference date: 21-06-2021 Through 25-06-2021",

year = "2021",

doi = "10.1007/978-3-030-78710-3_61",

language = "English",

isbn = "9783030787097",

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

publisher = "Springer",

pages = "639--649",

editor = "Ennis, {Daniel B.} and Perotti, {Luigi E.} and Wang, {Vicky Y.}",

booktitle = "Functional Imaging and Modeling of the Heart",

url = "https://fimh2021.github.io/",

}

Morales, X, Mill, J, Simeon, G, Juhl, KA, De Backer, O, Paulsen, RR & Camara, O 2021, Geometric Deep Learning for the Assessment of Thrombosis Risk in the Left Atrial Appendage. in DB Ennis, LE Perotti & VY Wang (eds), Functional Imaging and Modeling of the Heart. Springer, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 12738 LNCS, pp. 639-649, 11^th International Conference on Functional Imaging and Modeling of the Heart, Virtual, Online, 21/06/2021. https://doi.org/10.1007/978-3-030-78710-3_61

Geometric Deep Learning for the Assessment of Thrombosis Risk in the Left Atrial Appendage. / Morales, Xabier; Mill, Jordi; Simeon, Guillem et al.
Functional Imaging and Modeling of the Heart. ed. / Daniel B. Ennis; Luigi E. Perotti; Vicky Y. Wang. Springer, 2021. p. 639-649 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 12738 LNCS).

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

TY - GEN

T1 - Geometric Deep Learning for the Assessment of Thrombosis Risk in the Left Atrial Appendage

AU - Morales, Xabier

AU - Mill, Jordi

AU - Simeon, Guillem

AU - Juhl, Kristine A.

AU - De Backer, Ole

AU - Paulsen, Rasmus R.

AU - Camara, Oscar

N1 - Funding Information: Funding. This work was supported by the Agency for Management of University and Research Grants of the Generalitat de Catalunya under the Grants for the Contracting of New Research Staff Programme - FI (2020-FI-B-00690) and the Spanish Ministry of Economy and Competitiveness under the Programme for the Formation of Doctors (PRE2018-084062), the Maria de Maeztu Units of Excellence Programme (MDM-2015-0502) and the Retos Investigación project (RTI2018-101193-B-I00). Additionally, this work was supported by the H2020 EU SimCardioTest project (Digital transformation in Health and Care SC1-DTH-06-2020; grant agreement No. 101016496). Publisher Copyright: © 2021, Springer Nature Switzerland AG.

PY - 2021

Y1 - 2021

N2 - The assessment of left atrial appendage (LAA) thrombogenesis has experienced major advances with the adoption of patient-specific computational fluid dynamics (CFD) simulations. Nonetheless, due to the vast computational resources and long execution times required by fluid dynamics solvers, there is an ever-growing body of work aiming to develop surrogate models of fluid flow simulations based on neural networks. The present study builds on this foundation by developing a deep learning (DL) framework capable of predicting the endothelial cell activation potential (ECAP), linked to the risk of thrombosis, solely from the patient-specific LAA geometry. To this end, we leveraged recent advancements in Geometric DL, which seamlessly extend the unparalleled potential of convolutional neural networks (CNN), to non-Euclidean data such as meshes. The model was trained with a dataset combining 202 synthetic and 54 real LAA, predicting the ECAP distributions instantaneously, with an average mean absolute error of 0.563. Moreover, the resulting framework manages to predict the anatomical features related to higher ECAP values even when trained exclusively on synthetic cases.

AB - The assessment of left atrial appendage (LAA) thrombogenesis has experienced major advances with the adoption of patient-specific computational fluid dynamics (CFD) simulations. Nonetheless, due to the vast computational resources and long execution times required by fluid dynamics solvers, there is an ever-growing body of work aiming to develop surrogate models of fluid flow simulations based on neural networks. The present study builds on this foundation by developing a deep learning (DL) framework capable of predicting the endothelial cell activation potential (ECAP), linked to the risk of thrombosis, solely from the patient-specific LAA geometry. To this end, we leveraged recent advancements in Geometric DL, which seamlessly extend the unparalleled potential of convolutional neural networks (CNN), to non-Euclidean data such as meshes. The model was trained with a dataset combining 202 synthetic and 54 real LAA, predicting the ECAP distributions instantaneously, with an average mean absolute error of 0.563. Moreover, the resulting framework manages to predict the anatomical features related to higher ECAP values even when trained exclusively on synthetic cases.

KW - Computational fluid dynamics

KW - Geometric deep learning

KW - Left atrial appendage

KW - Thrombus formation

U2 - 10.1007/978-3-030-78710-3_61

DO - 10.1007/978-3-030-78710-3_61

M3 - Article in proceedings

AN - SCOPUS:85111857887

SN - 9783030787097

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

SP - 639

EP - 649

BT - Functional Imaging and Modeling of the Heart

A2 - Ennis, Daniel B.

A2 - Perotti, Luigi E.

A2 - Wang, Vicky Y.

PB - Springer

T2 - 11<sup>th</sup> International Conference on Functional Imaging and Modeling of the Heart

Y2 - 21 June 2021 through 25 June 2021

ER -

Morales X, Mill J, Simeon G, Juhl KA, De Backer O, Paulsen RR et al. Geometric Deep Learning for the Assessment of Thrombosis Risk in the Left Atrial Appendage. In Ennis DB, Perotti LE, Wang VY, editors, Functional Imaging and Modeling of the Heart. Springer. 2021. p. 639-649. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 12738 LNCS). doi: 10.1007/978-3-030-78710-3_61

Geometric Deep Learning for the Assessment of Thrombosis Risk in the Left Atrial Appendage

Abstract

Conference

Bibliographical note

Keywords

Access to Document

OpenUrl availability

Fingerprint

Cite this