Image Registration via Stochastic Gradient Markov Chain Monte Carlo

Daniel Grzech; Bernhard Kainz; Ben Glocker; Loïc le Folgoc

doi:10.1007/978-3-030-60365-6_1

Image Registration via Stochastic Gradient Markov Chain Monte Carlo

Daniel Grzech, Bernhard Kainz, Ben Glocker, Loïc le Folgoc

Imperial College London

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

Abstract

We develop a fully Bayesian framework for non-rigid registration of three-dimensional medical images, with a focus on uncertainty quantification. Probabilistic registration of large images along with calibrated uncertainty estimates is difficult for both computational and modelling reasons. To address the computational issues, we explore connections between the Markov chain Monte Carlo by backprop and the variational inference by backprop frameworks in order to efficiently draw thousands of samples from the posterior distribution. Regarding the modelling issues, we carefully design a Bayesian model for registration to overcome the existing barriers when using a dense, high-dimensional, and diffeomorphic parameterisation of the transformation. This results in improved calibration of uncertainty estimates.

Original language	English
Title of host publication	Uncertainty for Safe Utilization of Machine Learning in Medical Imaging, and Graphs in Biomedical Image Analysis - 2nd International Workshop, UNSURE 2020, and 3rd International Workshop, GRAIL 2020, Held in Conjunction with MICCAI 2020, Proceedings
Editors	Carole H. Sudre, Hamid Fehri, Tal Arbel, Christian F. Baumgartner, Adrian Dalca, Ryutaro Tanno, Koen Van Leemput, William M. Wells, Aristeidis Sotiras, Bartlomiej Papiez, Enzo Ferrante, Sarah Parisot
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	3-12
Number of pages	10
ISBN (Print)	9783030603649
DOIs	https://doi.org/10.1007/978-3-030-60365-6_1
Publication status	Published - 1 Jan 2020
Externally published	Yes
Event	2nd International Workshop on Uncertainty for Safe Utilization of Machine Learning in Medical Imaging, UNSURE 2020, and the 3rd International Workshop on Graphs in Biomedical Image Analysis, GRAIL 2020, held in conjunction with the 23rd International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2020 - Lima, Peru Duration: 8 Oct 2020 → 8 Oct 2020

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	12443 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	2nd International Workshop on Uncertainty for Safe Utilization of Machine Learning in Medical Imaging, UNSURE 2020, and the 3rd International Workshop on Graphs in Biomedical Image Analysis, GRAIL 2020, held in conjunction with the 23rd International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2020
Country/Territory	Peru
City	Lima
Period	8/10/20 → 8/10/20

Access to Document

10.1007/978-3-030-60365-6_1

Cite this

Grzech, D., Kainz, B., Glocker, B., & le Folgoc, L. (2020). Image Registration via Stochastic Gradient Markov Chain Monte Carlo. In C. H. Sudre, H. Fehri, T. Arbel, C. F. Baumgartner, A. Dalca, R. Tanno, K. Van Leemput, W. M. Wells, A. Sotiras, B. Papiez, E. Ferrante, & S. Parisot (Eds.), Uncertainty for Safe Utilization of Machine Learning in Medical Imaging, and Graphs in Biomedical Image Analysis - 2nd International Workshop, UNSURE 2020, and 3rd International Workshop, GRAIL 2020, Held in Conjunction with MICCAI 2020, Proceedings (pp. 3-12). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12443 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-030-60365-6_1

Grzech, Daniel ; Kainz, Bernhard ; Glocker, Ben et al. / Image Registration via Stochastic Gradient Markov Chain Monte Carlo. Uncertainty for Safe Utilization of Machine Learning in Medical Imaging, and Graphs in Biomedical Image Analysis - 2nd International Workshop, UNSURE 2020, and 3rd International Workshop, GRAIL 2020, Held in Conjunction with MICCAI 2020, Proceedings. editor / Carole H. Sudre ; Hamid Fehri ; Tal Arbel ; Christian F. Baumgartner ; Adrian Dalca ; Ryutaro Tanno ; Koen Van Leemput ; William M. Wells ; Aristeidis Sotiras ; Bartlomiej Papiez ; Enzo Ferrante ; Sarah Parisot. Springer Science and Business Media Deutschland GmbH, 2020. pp. 3-12 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{a70c015926184caf95ad60ae0f24782c,

title = "Image Registration via Stochastic Gradient Markov Chain Monte Carlo",

abstract = "We develop a fully Bayesian framework for non-rigid registration of three-dimensional medical images, with a focus on uncertainty quantification. Probabilistic registration of large images along with calibrated uncertainty estimates is difficult for both computational and modelling reasons. To address the computational issues, we explore connections between the Markov chain Monte Carlo by backprop and the variational inference by backprop frameworks in order to efficiently draw thousands of samples from the posterior distribution. Regarding the modelling issues, we carefully design a Bayesian model for registration to overcome the existing barriers when using a dense, high-dimensional, and diffeomorphic parameterisation of the transformation. This results in improved calibration of uncertainty estimates.",

author = "Daniel Grzech and Bernhard Kainz and Ben Glocker and \{le Folgoc\}, Lo{\"i}c",

note = "Publisher Copyright: {\textcopyright} 2020, Springer Nature Switzerland AG.; 2nd International Workshop on Uncertainty for Safe Utilization of Machine Learning in Medical Imaging, UNSURE 2020, and the 3rd International Workshop on Graphs in Biomedical Image Analysis, GRAIL 2020, held in conjunction with the 23rd International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2020 ; Conference date: 08-10-2020 Through 08-10-2020",

year = "2020",

month = jan,

day = "1",

doi = "10.1007/978-3-030-60365-6\_1",

language = "English",

isbn = "9783030603649",

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 = "3--12",

editor = "Sudre, \{Carole H.\} and Hamid Fehri and Tal Arbel and Baumgartner, \{Christian F.\} and Adrian Dalca and Ryutaro Tanno and \{Van Leemput\}, Koen and Wells, \{William M.\} and Aristeidis Sotiras and Bartlomiej Papiez and Enzo Ferrante and Sarah Parisot",

booktitle = "Uncertainty for Safe Utilization of Machine Learning in Medical Imaging, and Graphs in Biomedical Image Analysis - 2nd International Workshop, UNSURE 2020, and 3rd International Workshop, GRAIL 2020, Held in Conjunction with MICCAI 2020, Proceedings",

}

Grzech, D, Kainz, B, Glocker, B & le Folgoc, L 2020, Image Registration via Stochastic Gradient Markov Chain Monte Carlo. in CH Sudre, H Fehri, T Arbel, CF Baumgartner, A Dalca, R Tanno, K Van Leemput, WM Wells, A Sotiras, B Papiez, E Ferrante & S Parisot (eds), Uncertainty for Safe Utilization of Machine Learning in Medical Imaging, and Graphs in Biomedical Image Analysis - 2nd International Workshop, UNSURE 2020, and 3rd International Workshop, GRAIL 2020, Held in Conjunction with MICCAI 2020, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 12443 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 3-12, 2nd International Workshop on Uncertainty for Safe Utilization of Machine Learning in Medical Imaging, UNSURE 2020, and the 3rd International Workshop on Graphs in Biomedical Image Analysis, GRAIL 2020, held in conjunction with the 23rd International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2020, Lima, Peru, 8/10/20. https://doi.org/10.1007/978-3-030-60365-6_1

Image Registration via Stochastic Gradient Markov Chain Monte Carlo. / Grzech, Daniel; Kainz, Bernhard; Glocker, Ben et al.
Uncertainty for Safe Utilization of Machine Learning in Medical Imaging, and Graphs in Biomedical Image Analysis - 2nd International Workshop, UNSURE 2020, and 3rd International Workshop, GRAIL 2020, Held in Conjunction with MICCAI 2020, Proceedings. ed. / Carole H. Sudre; Hamid Fehri; Tal Arbel; Christian F. Baumgartner; Adrian Dalca; Ryutaro Tanno; Koen Van Leemput; William M. Wells; Aristeidis Sotiras; Bartlomiej Papiez; Enzo Ferrante; Sarah Parisot. Springer Science and Business Media Deutschland GmbH, 2020. p. 3-12 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12443 LNCS).

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

TY - GEN

T1 - Image Registration via Stochastic Gradient Markov Chain Monte Carlo

AU - Grzech, Daniel

AU - Kainz, Bernhard

AU - Glocker, Ben

AU - le Folgoc, Loïc

PY - 2020/1/1

Y1 - 2020/1/1

N2 - We develop a fully Bayesian framework for non-rigid registration of three-dimensional medical images, with a focus on uncertainty quantification. Probabilistic registration of large images along with calibrated uncertainty estimates is difficult for both computational and modelling reasons. To address the computational issues, we explore connections between the Markov chain Monte Carlo by backprop and the variational inference by backprop frameworks in order to efficiently draw thousands of samples from the posterior distribution. Regarding the modelling issues, we carefully design a Bayesian model for registration to overcome the existing barriers when using a dense, high-dimensional, and diffeomorphic parameterisation of the transformation. This results in improved calibration of uncertainty estimates.

AB - We develop a fully Bayesian framework for non-rigid registration of three-dimensional medical images, with a focus on uncertainty quantification. Probabilistic registration of large images along with calibrated uncertainty estimates is difficult for both computational and modelling reasons. To address the computational issues, we explore connections between the Markov chain Monte Carlo by backprop and the variational inference by backprop frameworks in order to efficiently draw thousands of samples from the posterior distribution. Regarding the modelling issues, we carefully design a Bayesian model for registration to overcome the existing barriers when using a dense, high-dimensional, and diffeomorphic parameterisation of the transformation. This results in improved calibration of uncertainty estimates.

U2 - 10.1007/978-3-030-60365-6_1

DO - 10.1007/978-3-030-60365-6_1

M3 - Conference contribution

AN - SCOPUS:85093074342

SN - 9783030603649

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

SP - 3

EP - 12

BT - Uncertainty for Safe Utilization of Machine Learning in Medical Imaging, and Graphs in Biomedical Image Analysis - 2nd International Workshop, UNSURE 2020, and 3rd International Workshop, GRAIL 2020, Held in Conjunction with MICCAI 2020, Proceedings

A2 - Sudre, Carole H.

A2 - Fehri, Hamid

A2 - Arbel, Tal

A2 - Baumgartner, Christian F.

A2 - Dalca, Adrian

A2 - Tanno, Ryutaro

A2 - Van Leemput, Koen

A2 - Wells, William M.

A2 - Sotiras, Aristeidis

A2 - Papiez, Bartlomiej

A2 - Ferrante, Enzo

A2 - Parisot, Sarah

PB - Springer Science and Business Media Deutschland GmbH

T2 - 2nd International Workshop on Uncertainty for Safe Utilization of Machine Learning in Medical Imaging, UNSURE 2020, and the 3rd International Workshop on Graphs in Biomedical Image Analysis, GRAIL 2020, held in conjunction with the 23rd International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2020

Y2 - 8 October 2020 through 8 October 2020

ER -

Grzech D, Kainz B, Glocker B, le Folgoc L. Image Registration via Stochastic Gradient Markov Chain Monte Carlo. In Sudre CH, Fehri H, Arbel T, Baumgartner CF, Dalca A, Tanno R, Van Leemput K, Wells WM, Sotiras A, Papiez B, Ferrante E, Parisot S, editors, Uncertainty for Safe Utilization of Machine Learning in Medical Imaging, and Graphs in Biomedical Image Analysis - 2nd International Workshop, UNSURE 2020, and 3rd International Workshop, GRAIL 2020, Held in Conjunction with MICCAI 2020, Proceedings. Springer Science and Business Media Deutschland GmbH. 2020. p. 3-12. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-60365-6_1

Image Registration via Stochastic Gradient Markov Chain Monte Carlo

Abstract

Publication series

Conference

Access to Document

Fingerprint

Cite this