MIMoSA: An Automated Method for Intermodal Segmentation Analysis of Multiple Sclerosis Brain Lesions

Alessandra M. Valcarcel, Kristin A. Linn, Simon N. Vandekar, Theodore D. Satterthwaite, John Muschelli, Peter Calabresi, Dzung L. Pham, Melissa Lynne Martin, Russell T. Shinohara

Research output: Contribution to journalArticle

Abstract

BACKGROUND AND PURPOSE: Magnetic resonance imaging (MRI) is crucial for in vivo detection and characterization of white matter lesions (WMLs) in multiple sclerosis. While WMLs have been studied for over two decades using MRI, automated segmentation remains challenging. Although the majority of statistical techniques for the automated segmentation of WMLs are based on single imaging modalities, recent advances have used multimodal techniques for identifying WMLs. Complementary modalities emphasize different tissue properties, which help identify interrelated features of lesions. METHODS: Method for Inter-Modal Segmentation Analysis (MIMoSA), a fully automatic lesion segmentation algorithm that utilizes novel covariance features from intermodal coupling regression in addition to mean structure to model the probability lesion is contained in each voxel, is proposed. MIMoSA was validated by comparison with both expert manual and other automated segmentation methods in two datasets. The first included 98 subjects imaged at Johns Hopkins Hospital in which bootstrap cross-validation was used to compare the performance of MIMoSA against OASIS and LesionTOADS, two popular automatic segmentation approaches. For a secondary validation, a publicly available data from a segmentation challenge were used for performance benchmarking. RESULTS: In the Johns Hopkins study, MIMoSA yielded average Sørensen-Dice coefficient (DSC) of .57 and partial AUC of .68 calculated with false positive rates up to 1%. This was superior to performance using OASIS and LesionTOADS. The proposed method also performed competitively in the segmentation challenge dataset. CONCLUSION: MIMoSA resulted in statistically significant improvements in lesion segmentation performance compared with LesionTOADS and OASIS, and performed competitively in an additional validation study.

Original languageEnglish (US)
JournalJournal of Neuroimaging
DOIs
StateAccepted/In press - Jan 1 2018

Fingerprint

Multiple Sclerosis
Brain
Magnetic Resonance Imaging
Benchmarking
Validation Studies
Area Under Curve
White Matter
poly(divinyl-co-N-vinylpyrrolidinone)

Keywords

  • Automatic segmentation
  • Lesion detection
  • Logistic regression
  • Multiple sclerosis

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging
  • Clinical Neurology

Cite this

MIMoSA : An Automated Method for Intermodal Segmentation Analysis of Multiple Sclerosis Brain Lesions. / Valcarcel, Alessandra M.; Linn, Kristin A.; Vandekar, Simon N.; Satterthwaite, Theodore D.; Muschelli, John; Calabresi, Peter; Pham, Dzung L.; Martin, Melissa Lynne; Shinohara, Russell T.

In: Journal of Neuroimaging, 01.01.2018.

Research output: Contribution to journalArticle

Valcarcel, Alessandra M. ; Linn, Kristin A. ; Vandekar, Simon N. ; Satterthwaite, Theodore D. ; Muschelli, John ; Calabresi, Peter ; Pham, Dzung L. ; Martin, Melissa Lynne ; Shinohara, Russell T. / MIMoSA : An Automated Method for Intermodal Segmentation Analysis of Multiple Sclerosis Brain Lesions. In: Journal of Neuroimaging. 2018.
@article{0a094776e7f242a3aec278f586f651f0,
title = "MIMoSA: An Automated Method for Intermodal Segmentation Analysis of Multiple Sclerosis Brain Lesions",
abstract = "BACKGROUND AND PURPOSE: Magnetic resonance imaging (MRI) is crucial for in vivo detection and characterization of white matter lesions (WMLs) in multiple sclerosis. While WMLs have been studied for over two decades using MRI, automated segmentation remains challenging. Although the majority of statistical techniques for the automated segmentation of WMLs are based on single imaging modalities, recent advances have used multimodal techniques for identifying WMLs. Complementary modalities emphasize different tissue properties, which help identify interrelated features of lesions. METHODS: Method for Inter-Modal Segmentation Analysis (MIMoSA), a fully automatic lesion segmentation algorithm that utilizes novel covariance features from intermodal coupling regression in addition to mean structure to model the probability lesion is contained in each voxel, is proposed. MIMoSA was validated by comparison with both expert manual and other automated segmentation methods in two datasets. The first included 98 subjects imaged at Johns Hopkins Hospital in which bootstrap cross-validation was used to compare the performance of MIMoSA against OASIS and LesionTOADS, two popular automatic segmentation approaches. For a secondary validation, a publicly available data from a segmentation challenge were used for performance benchmarking. RESULTS: In the Johns Hopkins study, MIMoSA yielded average S{\o}rensen-Dice coefficient (DSC) of .57 and partial AUC of .68 calculated with false positive rates up to 1{\%}. This was superior to performance using OASIS and LesionTOADS. The proposed method also performed competitively in the segmentation challenge dataset. CONCLUSION: MIMoSA resulted in statistically significant improvements in lesion segmentation performance compared with LesionTOADS and OASIS, and performed competitively in an additional validation study.",
keywords = "Automatic segmentation, Lesion detection, Logistic regression, Multiple sclerosis",
author = "Valcarcel, {Alessandra M.} and Linn, {Kristin A.} and Vandekar, {Simon N.} and Satterthwaite, {Theodore D.} and John Muschelli and Peter Calabresi and Pham, {Dzung L.} and Martin, {Melissa Lynne} and Shinohara, {Russell T.}",
year = "2018",
month = "1",
day = "1",
doi = "10.1111/jon.12506",
language = "English (US)",
journal = "Journal of Neuroimaging",
issn = "1051-2284",
publisher = "Wiley-Blackwell",

}

TY - JOUR

T1 - MIMoSA

T2 - An Automated Method for Intermodal Segmentation Analysis of Multiple Sclerosis Brain Lesions

AU - Valcarcel, Alessandra M.

AU - Linn, Kristin A.

AU - Vandekar, Simon N.

AU - Satterthwaite, Theodore D.

AU - Muschelli, John

AU - Calabresi, Peter

AU - Pham, Dzung L.

AU - Martin, Melissa Lynne

AU - Shinohara, Russell T.

PY - 2018/1/1

Y1 - 2018/1/1

N2 - BACKGROUND AND PURPOSE: Magnetic resonance imaging (MRI) is crucial for in vivo detection and characterization of white matter lesions (WMLs) in multiple sclerosis. While WMLs have been studied for over two decades using MRI, automated segmentation remains challenging. Although the majority of statistical techniques for the automated segmentation of WMLs are based on single imaging modalities, recent advances have used multimodal techniques for identifying WMLs. Complementary modalities emphasize different tissue properties, which help identify interrelated features of lesions. METHODS: Method for Inter-Modal Segmentation Analysis (MIMoSA), a fully automatic lesion segmentation algorithm that utilizes novel covariance features from intermodal coupling regression in addition to mean structure to model the probability lesion is contained in each voxel, is proposed. MIMoSA was validated by comparison with both expert manual and other automated segmentation methods in two datasets. The first included 98 subjects imaged at Johns Hopkins Hospital in which bootstrap cross-validation was used to compare the performance of MIMoSA against OASIS and LesionTOADS, two popular automatic segmentation approaches. For a secondary validation, a publicly available data from a segmentation challenge were used for performance benchmarking. RESULTS: In the Johns Hopkins study, MIMoSA yielded average Sørensen-Dice coefficient (DSC) of .57 and partial AUC of .68 calculated with false positive rates up to 1%. This was superior to performance using OASIS and LesionTOADS. The proposed method also performed competitively in the segmentation challenge dataset. CONCLUSION: MIMoSA resulted in statistically significant improvements in lesion segmentation performance compared with LesionTOADS and OASIS, and performed competitively in an additional validation study.

AB - BACKGROUND AND PURPOSE: Magnetic resonance imaging (MRI) is crucial for in vivo detection and characterization of white matter lesions (WMLs) in multiple sclerosis. While WMLs have been studied for over two decades using MRI, automated segmentation remains challenging. Although the majority of statistical techniques for the automated segmentation of WMLs are based on single imaging modalities, recent advances have used multimodal techniques for identifying WMLs. Complementary modalities emphasize different tissue properties, which help identify interrelated features of lesions. METHODS: Method for Inter-Modal Segmentation Analysis (MIMoSA), a fully automatic lesion segmentation algorithm that utilizes novel covariance features from intermodal coupling regression in addition to mean structure to model the probability lesion is contained in each voxel, is proposed. MIMoSA was validated by comparison with both expert manual and other automated segmentation methods in two datasets. The first included 98 subjects imaged at Johns Hopkins Hospital in which bootstrap cross-validation was used to compare the performance of MIMoSA against OASIS and LesionTOADS, two popular automatic segmentation approaches. For a secondary validation, a publicly available data from a segmentation challenge were used for performance benchmarking. RESULTS: In the Johns Hopkins study, MIMoSA yielded average Sørensen-Dice coefficient (DSC) of .57 and partial AUC of .68 calculated with false positive rates up to 1%. This was superior to performance using OASIS and LesionTOADS. The proposed method also performed competitively in the segmentation challenge dataset. CONCLUSION: MIMoSA resulted in statistically significant improvements in lesion segmentation performance compared with LesionTOADS and OASIS, and performed competitively in an additional validation study.

KW - Automatic segmentation

KW - Lesion detection

KW - Logistic regression

KW - Multiple sclerosis

UR - http://www.scopus.com/inward/record.url?scp=85043338865&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85043338865&partnerID=8YFLogxK

U2 - 10.1111/jon.12506

DO - 10.1111/jon.12506

M3 - Article

C2 - 29516669

AN - SCOPUS:85043338865

JO - Journal of Neuroimaging

JF - Journal of Neuroimaging

SN - 1051-2284

ER -