N3 Bias Field Correction Explained as a Bayesian Modeling Method

Christian Thode Larsen; Juan Eugenio Iglesias; Koen Van Leemput

doi:10.1007/978-3-319-12289-2_1

N3 Bias Field Correction Explained as a Bayesian Modeling Method

Christian Thode Larsen, Juan Eugenio Iglesias, Koen Van Leemput

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

Abstract

Although N3 is perhaps the most widely used method for MRI bias field correction, its underlying mechanism is in fact not well understood. Specifically, the method relies on a relatively heuristic recipe of alternating iterative steps that does not optimize any particular objective function. In this paper we explain the successful bias field correction properties of N3 by showing that it implicitly uses the same generative models and computational strategies as expectation maximization (EM) based bias field correction methods. We demonstrate experimentally that purely EM-based methods are capable of producing bias field correction results comparable to those of N3 in less computation time.

Original language	English
Title of host publication	Bayesian and Graphical Models for Biomedical Imaging : Revised Selected Papers of the first International Workshop on Bayesian and Grahical Models for Biomedical Imaging, BAMBI 2014
Publisher	Springer
Publication date	2014
Pages	1-12
ISBN (Print)	978-3-319-12288-5
ISBN (Electronic)	978-3-319-12289-2
DOIs	https://doi.org/10.1007/978-3-319-12289-2_1
Publication status	Published - 2014
Event	1st International Workshop on Bayesian and Graphical Models for Biomedical Imaging, BAMBI 2014 - Cambridge, United States Duration: 18 Sept 2014 → … Conference number: 1 http://bambi.cs.ucl.ac.uk/

Workshop

Workshop	1st International Workshop on Bayesian and Graphical Models for Biomedical Imaging, BAMBI 2014
Number	1
Country/Territory	United States
City	Cambridge
Period	18/09/2014 → …
Other	In correlation with the 17th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI)
Internet address	http://bambi.cs.ucl.ac.uk/

Series	Lecture Notes in Computer Science
Number	8677
ISSN	0302-9743

Access to Document

10.1007/978-3-319-12289-2_1

OpenUrl availability

Full text

Cite this

Larsen, C. T., Iglesias, J. E., & Van Leemput, K. (2014). N3 Bias Field Correction Explained as a Bayesian Modeling Method. In Bayesian and Graphical Models for Biomedical Imaging: Revised Selected Papers of the first International Workshop on Bayesian and Grahical Models for Biomedical Imaging, BAMBI 2014 (pp. 1-12). Springer. https://doi.org/10.1007/978-3-319-12289-2_1

Larsen, Christian Thode ; Iglesias, Juan Eugenio ; Van Leemput, Koen. / N3 Bias Field Correction Explained as a Bayesian Modeling Method. Bayesian and Graphical Models for Biomedical Imaging: Revised Selected Papers of the first International Workshop on Bayesian and Grahical Models for Biomedical Imaging, BAMBI 2014. Springer, 2014. pp. 1-12 (Lecture Notes in Computer Science; No. 8677).

@inproceedings{fe0cb37946d14b979100cb9c81f77f07,

title = "N3 Bias Field Correction Explained as a Bayesian Modeling Method",

abstract = "Although N3 is perhaps the most widely used method for MRI bias field correction, its underlying mechanism is in fact not well understood. Specifically, the method relies on a relatively heuristic recipe of alternating iterative steps that does not optimize any particular objective function. In this paper we explain the successful bias field correction properties of N3 by showing that it implicitly uses the same generative models and computational strategies as expectation maximization (EM) based bias field correction methods. We demonstrate experimentally that purely EM-based methods are capable of producing bias field correction results comparable to those of N3 in less computation time.",

author = "Larsen, {Christian Thode} and Iglesias, {Juan Eugenio} and {Van Leemput}, Koen",

year = "2014",

doi = "10.1007/978-3-319-12289-2_1",

language = "English",

isbn = "978-3-319-12288-5",

series = "Lecture Notes in Computer Science",

publisher = "Springer",

number = "8677",

pages = "1--12",

booktitle = "Bayesian and Graphical Models for Biomedical Imaging",

note = "1st International Workshop on Bayesian and Graphical Models for Biomedical Imaging, BAMBI 2014, BAMBI ; Conference date: 18-09-2014",

url = "http://bambi.cs.ucl.ac.uk/",

}

Larsen, CT, Iglesias, JE & Van Leemput, K 2014, N3 Bias Field Correction Explained as a Bayesian Modeling Method. in Bayesian and Graphical Models for Biomedical Imaging: Revised Selected Papers of the first International Workshop on Bayesian and Grahical Models for Biomedical Imaging, BAMBI 2014. Springer, Lecture Notes in Computer Science, no. 8677, pp. 1-12, 1st International Workshop on Bayesian and Graphical Models for Biomedical Imaging, BAMBI 2014, Cambridge, United States, 18/09/2014. https://doi.org/10.1007/978-3-319-12289-2_1

N3 Bias Field Correction Explained as a Bayesian Modeling Method. / Larsen, Christian Thode; Iglesias, Juan Eugenio; Van Leemput, Koen.
Bayesian and Graphical Models for Biomedical Imaging: Revised Selected Papers of the first International Workshop on Bayesian and Grahical Models for Biomedical Imaging, BAMBI 2014. Springer, 2014. p. 1-12 (Lecture Notes in Computer Science; No. 8677).

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

TY - GEN

T1 - N3 Bias Field Correction Explained as a Bayesian Modeling Method

AU - Larsen, Christian Thode

AU - Iglesias, Juan Eugenio

AU - Van Leemput, Koen

N1 - Conference code: 1

PY - 2014

Y1 - 2014

N2 - Although N3 is perhaps the most widely used method for MRI bias field correction, its underlying mechanism is in fact not well understood. Specifically, the method relies on a relatively heuristic recipe of alternating iterative steps that does not optimize any particular objective function. In this paper we explain the successful bias field correction properties of N3 by showing that it implicitly uses the same generative models and computational strategies as expectation maximization (EM) based bias field correction methods. We demonstrate experimentally that purely EM-based methods are capable of producing bias field correction results comparable to those of N3 in less computation time.

AB - Although N3 is perhaps the most widely used method for MRI bias field correction, its underlying mechanism is in fact not well understood. Specifically, the method relies on a relatively heuristic recipe of alternating iterative steps that does not optimize any particular objective function. In this paper we explain the successful bias field correction properties of N3 by showing that it implicitly uses the same generative models and computational strategies as expectation maximization (EM) based bias field correction methods. We demonstrate experimentally that purely EM-based methods are capable of producing bias field correction results comparable to those of N3 in less computation time.

U2 - 10.1007/978-3-319-12289-2_1

DO - 10.1007/978-3-319-12289-2_1

M3 - Article in proceedings

SN - 978-3-319-12288-5

T3 - Lecture Notes in Computer Science

SP - 1

EP - 12

BT - Bayesian and Graphical Models for Biomedical Imaging

PB - Springer

T2 - 1st International Workshop on Bayesian and Graphical Models for Biomedical Imaging, BAMBI 2014

Y2 - 18 September 2014

ER -

Larsen CT, Iglesias JE, Van Leemput K. N3 Bias Field Correction Explained as a Bayesian Modeling Method. In Bayesian and Graphical Models for Biomedical Imaging: Revised Selected Papers of the first International Workshop on Bayesian and Grahical Models for Biomedical Imaging, BAMBI 2014. Springer. 2014. p. 1-12. (Lecture Notes in Computer Science; No. 8677). doi: 10.1007/978-3-319-12289-2_1

N3 Bias Field Correction Explained as a Bayesian Modeling Method

Abstract

Workshop

Access to Document

OpenUrl availability

Fingerprint

Cite this