Transfer learning from simulations on a reference anatomy for ECGI in personalized cardiac resynchronization therapy

Sophie Giffard-Roisin; Herve Delingette; Thomas Jackson; Jessica Webb; Lauren Fovargue; Jack Lee; C. Aldo Rinaldi; Reza Razavi; Nicholas Ayache; Maxime Sermesant

doi:10.1109/TBME.2018.2839713

Transfer learning from simulations on a reference anatomy for ECGI in personalized cardiac resynchronization therapy

Sophie Giffard-Roisin, Herve Delingette, Thomas Jackson, Jessica Webb, Lauren Fovargue, Jack Lee, C. Aldo Rinaldi, Reza Razavi, Nicholas Ayache, Maxime Sermesant

Inria French Institute for Research in Computer Science and Automation

Research output: Contribution to journal › Article › peer-review

30 Citations (Scopus)

150 Downloads (Pure)

Abstract

Goal: Noninvasive cardiac electrophysiology (EP) model personalisation has raised interest for instance in the scope of predicting EP cardiac resynchronization therapy (CRT) response. However, the restricted clinical applicability of current methods is due in particular to the limitation to simple situations and the important computational cost. Methods: We propose in this manuscript an approach to tackle these two issues. First, we analyze more complex propagation patterns (multiple onsets and scar tissue) using relevance vector regression and shape dimensionality reduction on a large simulated database. Second, this learning is performed offline on a reference anatomy and transferred onto patient-specific anatomies in order to achieve fast personalized predictions online. Results: We evaluated our method on a dataset composed of 20 dyssynchrony patients with a total of 120 different cardiac cycles. The comparison with a commercially available electrocardiographic imaging (ECGI) method shows a good identification of the cardiac activation pattern. From the cardiac parameters estimated in sinus rhythm, we predicted five different paced patterns for each patient. The comparison with the body surface potential mappings (BSPM) measured during pacing and the ECGI method indicates a good predictive power. Conclusion: We showed that learning offline from a large simulated database on a reference anatomy was able to capture the main cardiac EP characteristics from noninvasive measurements for fast patient-specific predictions. Significance: The fast CRT pacing predictions are a step forward to a noninvasive CRT patient selection and therapy optimisation, to help clinicians in these difficult tasks.

Original language	English
Article number	8362988
Pages (from-to)	343-353
Number of pages	11
Journal	IEEE Transactions on Biomedical Engineering
Volume	66
Issue number	2
Early online date	23 May 2018
DOIs	https://doi.org/10.1109/TBME.2018.2839713
Publication status	Published - Feb 2019

Keywords

Cardiac electrophysiology
ECG imaging
inverse problem of ECG
personalisation

Access to Document

10.1109/TBME.2018.2839713

Transfer Learning From Simulations_GIFFARD-ROISIN_Accepted22May2018Publishedonline23May2018_GREEN AAM
© 2019 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
Accepted author manuscript, 7.43 MB

https://hal.archives-ouvertes.fr/hal-01796483v2

Cite this

Giffard-Roisin, S., Delingette, H., Jackson, T., Webb, J., Fovargue, L., Lee, J., Rinaldi, C. A., Razavi, R., Ayache, N., & Sermesant, M. (2019). Transfer learning from simulations on a reference anatomy for ECGI in personalized cardiac resynchronization therapy. IEEE Transactions on Biomedical Engineering, 66(2), 343-353. Article 8362988. https://doi.org/10.1109/TBME.2018.2839713

@article{184359eb01874104b8e5a48d9e26a98f,

title = "Transfer learning from simulations on a reference anatomy for ECGI in personalized cardiac resynchronization therapy",

abstract = "Goal: Noninvasive cardiac electrophysiology (EP) model personalisation has raised interest for instance in the scope of predicting EP cardiac resynchronization therapy (CRT) response. However, the restricted clinical applicability of current methods is due in particular to the limitation to simple situations and the important computational cost. Methods: We propose in this manuscript an approach to tackle these two issues. First, we analyze more complex propagation patterns (multiple onsets and scar tissue) using relevance vector regression and shape dimensionality reduction on a large simulated database. Second, this learning is performed offline on a reference anatomy and transferred onto patient-specific anatomies in order to achieve fast personalized predictions online. Results: We evaluated our method on a dataset composed of 20 dyssynchrony patients with a total of 120 different cardiac cycles. The comparison with a commercially available electrocardiographic imaging (ECGI) method shows a good identification of the cardiac activation pattern. From the cardiac parameters estimated in sinus rhythm, we predicted five different paced patterns for each patient. The comparison with the body surface potential mappings (BSPM) measured during pacing and the ECGI method indicates a good predictive power. Conclusion: We showed that learning offline from a large simulated database on a reference anatomy was able to capture the main cardiac EP characteristics from noninvasive measurements for fast patient-specific predictions. Significance: The fast CRT pacing predictions are a step forward to a noninvasive CRT patient selection and therapy optimisation, to help clinicians in these difficult tasks.",

keywords = "Cardiac electrophysiology, ECG imaging, inverse problem of ECG, personalisation",

author = "Sophie Giffard-Roisin and Herve Delingette and Thomas Jackson and Jessica Webb and Lauren Fovargue and Jack Lee and Rinaldi, {C. Aldo} and Reza Razavi and Nicholas Ayache and Maxime Sermesant",

year = "2019",

month = feb,

doi = "10.1109/TBME.2018.2839713",

language = "English",

volume = "66",

pages = "343--353",

journal = "IEEE Transactions on Biomedical Engineering",

issn = "0018-9294",

publisher = "IEEE Computer Society",

number = "2",

}

Giffard-Roisin, S, Delingette, H, Jackson, T , Webb, J , Fovargue, L , Lee, J , Rinaldi, CA , Razavi, R, Ayache, N & Sermesant, M 2019, 'Transfer learning from simulations on a reference anatomy for ECGI in personalized cardiac resynchronization therapy', IEEE Transactions on Biomedical Engineering, vol. 66, no. 2, 8362988, pp. 343-353. https://doi.org/10.1109/TBME.2018.2839713

TY - JOUR

T1 - Transfer learning from simulations on a reference anatomy for ECGI in personalized cardiac resynchronization therapy

AU - Giffard-Roisin, Sophie

AU - Delingette, Herve

AU - Jackson, Thomas

AU - Webb, Jessica

AU - Fovargue, Lauren

AU - Lee, Jack

AU - Rinaldi, C. Aldo

AU - Razavi, Reza

AU - Ayache, Nicholas

AU - Sermesant, Maxime

PY - 2019/2

Y1 - 2019/2

N2 - Goal: Noninvasive cardiac electrophysiology (EP) model personalisation has raised interest for instance in the scope of predicting EP cardiac resynchronization therapy (CRT) response. However, the restricted clinical applicability of current methods is due in particular to the limitation to simple situations and the important computational cost. Methods: We propose in this manuscript an approach to tackle these two issues. First, we analyze more complex propagation patterns (multiple onsets and scar tissue) using relevance vector regression and shape dimensionality reduction on a large simulated database. Second, this learning is performed offline on a reference anatomy and transferred onto patient-specific anatomies in order to achieve fast personalized predictions online. Results: We evaluated our method on a dataset composed of 20 dyssynchrony patients with a total of 120 different cardiac cycles. The comparison with a commercially available electrocardiographic imaging (ECGI) method shows a good identification of the cardiac activation pattern. From the cardiac parameters estimated in sinus rhythm, we predicted five different paced patterns for each patient. The comparison with the body surface potential mappings (BSPM) measured during pacing and the ECGI method indicates a good predictive power. Conclusion: We showed that learning offline from a large simulated database on a reference anatomy was able to capture the main cardiac EP characteristics from noninvasive measurements for fast patient-specific predictions. Significance: The fast CRT pacing predictions are a step forward to a noninvasive CRT patient selection and therapy optimisation, to help clinicians in these difficult tasks.

AB - Goal: Noninvasive cardiac electrophysiology (EP) model personalisation has raised interest for instance in the scope of predicting EP cardiac resynchronization therapy (CRT) response. However, the restricted clinical applicability of current methods is due in particular to the limitation to simple situations and the important computational cost. Methods: We propose in this manuscript an approach to tackle these two issues. First, we analyze more complex propagation patterns (multiple onsets and scar tissue) using relevance vector regression and shape dimensionality reduction on a large simulated database. Second, this learning is performed offline on a reference anatomy and transferred onto patient-specific anatomies in order to achieve fast personalized predictions online. Results: We evaluated our method on a dataset composed of 20 dyssynchrony patients with a total of 120 different cardiac cycles. The comparison with a commercially available electrocardiographic imaging (ECGI) method shows a good identification of the cardiac activation pattern. From the cardiac parameters estimated in sinus rhythm, we predicted five different paced patterns for each patient. The comparison with the body surface potential mappings (BSPM) measured during pacing and the ECGI method indicates a good predictive power. Conclusion: We showed that learning offline from a large simulated database on a reference anatomy was able to capture the main cardiac EP characteristics from noninvasive measurements for fast patient-specific predictions. Significance: The fast CRT pacing predictions are a step forward to a noninvasive CRT patient selection and therapy optimisation, to help clinicians in these difficult tasks.

KW - Cardiac electrophysiology

KW - ECG imaging

KW - inverse problem of ECG

KW - personalisation

UR - http://www.scopus.com/inward/record.url?scp=85047622620&partnerID=8YFLogxK

U2 - 10.1109/TBME.2018.2839713

DO - 10.1109/TBME.2018.2839713

M3 - Article

AN - SCOPUS:85047622620

SN - 0018-9294

VL - 66

SP - 343

EP - 353

JO - IEEE Transactions on Biomedical Engineering

JF - IEEE Transactions on Biomedical Engineering

IS - 2

M1 - 8362988

ER -

Transfer learning from simulations on a reference anatomy for ECGI in personalized cardiac resynchronization therapy

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this