Bayesian Model Selection for Pathological Neuroimaging Data Applied to White Matter Lesion Segmentation

Carole H. Sudre; M. Jorge Cardoso; Willem H. Bouvy; Geert Jan Biessels; Josephine Barnes; Sebastien Ourselin

doi:10.1109/TMI.2015.2419072

Bayesian Model Selection for Pathological Neuroimaging Data Applied to White Matter Lesion Segmentation

Carole H. Sudre^*, M. Jorge Cardoso, Willem H. Bouvy, Geert Jan Biessels, Josephine Barnes, Sebastien Ourselin

^*Corresponding author for this work

Imaging and Biomedical Engineering

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

124 Citations (Scopus)

Abstract

In neuroimaging studies, pathologies can present themselves as abnormal intensity patterns. Thus, solutions for detecting abnormal intensities are currently under investigation. As each patient is unique, an unbiased and biologically plausible model of pathological data would have to be able to adapt to the subject's individual presentation. Such a model would provide the means for a better understanding of the underlying biological processes and improve one's ability to define pathologically meaningful imaging biomarkers. With this aim in mind, this work proposes a hierarchical fully unsupervised model selection framework for neuroimaging data which enables the distinction between different types of abnormal image patterns without pathological a priori knowledge. Its application on simulated and clinical data demonstrated the ability to detect abnormal intensity clusters, resulting in a competitive to improved behavior in white matter lesion segmentation when compared to three other freely-available automated methods.

Original language	English
Article number	7078891
Pages (from-to)	2079-2102
Number of pages	24
Journal	IEEE Transactions on Medical Imaging
Volume	34
Issue number	10
DOIs	https://doi.org/10.1109/TMI.2015.2419072
Publication status	Published - 1 Oct 2015

Keywords

Bayesian inference criterion (BIC)
brain segmentation
Gaussian mixture model (GMM)
magnetic resonance imaging (MRI)
split-and-merge (SM) strategy
white matter lesion (WML)

Access to Document

10.1109/TMI.2015.2419072Licence: CC BY

Cite this

@article{44c7039ba3a04742808283845e7fb902,

title = "Bayesian Model Selection for Pathological Neuroimaging Data Applied to White Matter Lesion Segmentation",

abstract = "In neuroimaging studies, pathologies can present themselves as abnormal intensity patterns. Thus, solutions for detecting abnormal intensities are currently under investigation. As each patient is unique, an unbiased and biologically plausible model of pathological data would have to be able to adapt to the subject's individual presentation. Such a model would provide the means for a better understanding of the underlying biological processes and improve one's ability to define pathologically meaningful imaging biomarkers. With this aim in mind, this work proposes a hierarchical fully unsupervised model selection framework for neuroimaging data which enables the distinction between different types of abnormal image patterns without pathological a priori knowledge. Its application on simulated and clinical data demonstrated the ability to detect abnormal intensity clusters, resulting in a competitive to improved behavior in white matter lesion segmentation when compared to three other freely-available automated methods.",

keywords = "Bayesian inference criterion (BIC), brain segmentation, Gaussian mixture model (GMM), magnetic resonance imaging (MRI), split-and-merge (SM) strategy, white matter lesion (WML)",

author = "Sudre, {Carole H.} and Cardoso, {M. Jorge} and Bouvy, {Willem H.} and Biessels, {Geert Jan} and Josephine Barnes and Sebastien Ourselin",

year = "2015",

month = oct,

day = "1",

doi = "10.1109/TMI.2015.2419072",

language = "English",

volume = "34",

pages = "2079--2102",

journal = "IEEE Transactions on Medical Imaging",

issn = "0278-0062",

publisher = "I E E E",

number = "10",

}

TY - JOUR

T1 - Bayesian Model Selection for Pathological Neuroimaging Data Applied to White Matter Lesion Segmentation

AU - Sudre, Carole H.

AU - Cardoso, M. Jorge

AU - Bouvy, Willem H.

AU - Biessels, Geert Jan

AU - Barnes, Josephine

AU - Ourselin, Sebastien

PY - 2015/10/1

Y1 - 2015/10/1

N2 - In neuroimaging studies, pathologies can present themselves as abnormal intensity patterns. Thus, solutions for detecting abnormal intensities are currently under investigation. As each patient is unique, an unbiased and biologically plausible model of pathological data would have to be able to adapt to the subject's individual presentation. Such a model would provide the means for a better understanding of the underlying biological processes and improve one's ability to define pathologically meaningful imaging biomarkers. With this aim in mind, this work proposes a hierarchical fully unsupervised model selection framework for neuroimaging data which enables the distinction between different types of abnormal image patterns without pathological a priori knowledge. Its application on simulated and clinical data demonstrated the ability to detect abnormal intensity clusters, resulting in a competitive to improved behavior in white matter lesion segmentation when compared to three other freely-available automated methods.

AB - In neuroimaging studies, pathologies can present themselves as abnormal intensity patterns. Thus, solutions for detecting abnormal intensities are currently under investigation. As each patient is unique, an unbiased and biologically plausible model of pathological data would have to be able to adapt to the subject's individual presentation. Such a model would provide the means for a better understanding of the underlying biological processes and improve one's ability to define pathologically meaningful imaging biomarkers. With this aim in mind, this work proposes a hierarchical fully unsupervised model selection framework for neuroimaging data which enables the distinction between different types of abnormal image patterns without pathological a priori knowledge. Its application on simulated and clinical data demonstrated the ability to detect abnormal intensity clusters, resulting in a competitive to improved behavior in white matter lesion segmentation when compared to three other freely-available automated methods.

KW - Bayesian inference criterion (BIC)

KW - brain segmentation

KW - Gaussian mixture model (GMM)

KW - magnetic resonance imaging (MRI)

KW - split-and-merge (SM) strategy

KW - white matter lesion (WML)

UR - http://www.scopus.com/inward/record.url?scp=84960478843&partnerID=8YFLogxK

U2 - 10.1109/TMI.2015.2419072

DO - 10.1109/TMI.2015.2419072

M3 - Article

C2 - 25850086

AN - SCOPUS:84960478843

SN - 0278-0062

VL - 34

SP - 2079

EP - 2102

JO - IEEE Transactions on Medical Imaging

JF - IEEE Transactions on Medical Imaging

IS - 10

M1 - 7078891

ER -

Bayesian Model Selection for Pathological Neuroimaging Data Applied to White Matter Lesion Segmentation

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this