MIMoSA: An Approach to automatically segment T2 hyperintense and T1 hypointense lesions in multiple sclerosis

Alessandra M. Valcarcel; Kristin A. Linn; Fariha Khalid; Simon N. Vandekar; Shahamat Tauhid; Theodore D. Satterthwaite; John Muschelli; Rohit Bakshi; Russell T. Shinohara

doi:10.1007/978-3-030-11723-8_5

MIMoSA: An Approach to automatically segment T2 hyperintense and T1 hypointense lesions in multiple sclerosis

Alessandra M. Valcarcel, Kristin A. Linn, Fariha Khalid, Simon N. Vandekar, Shahamat Tauhid, Theodore D. Satterthwaite, John Muschelli, Rohit Bakshi, Russell T. Shinohara

Bloomberg School of Public Health

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

Abstract

Magnetic resonance imaging (MRI) is crucial for in vivo detection and characterization of white matter lesions (WML) in multiple sclerosis (MS). The most widely established MRI outcome measure is the volume of hyperintense lesions on T2-weighted images (T2L). Unfortunately, T2L are non-specific for the level of tissue destruction and show a weak relationship to clinical status. Interest in lesions appearing hypointense on T1-weighted images (T1L) (“black holes”), which provide more specificity for axonal loss and a closer link to neurologic disability, has thus grown. The technical difficulty of T1L segmentation has led investigators to rely on time-consuming manual assessments prone to inter- and intra-rater variability. We implement MIMoSA, a current T2L automatic segmentation approach, to delineate T1L. Using cross-validation, MIMoSA proved robust for segmenting both T2L and T1L. For T2L, a Sørensen-Dice coefficient (DSC) of 0.6 and partial AUC (pAUC) up to 1% false positive rate of 0.69 were achieved. For T1L, 0.48 DSC and 0.63 pAUC were achieved. The correlation between EDSS and manual versus automatic volumes were similar for T1L (0.32 manual vs. 0.34 MIMoSA) and T2L (0.34 vs. 0.34).

Original language	English (US)
Title of host publication	Brainlesion
Subtitle of host publication	Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 4th International Workshop, BrainLes 2018, Held in Conjunction with MICCAI 2018, Revised Selected Papers
Editors	Farahani Keyvan, Spyridon Bakas, Theo van Walsum, Hugo Kuijf, Alessandro Crimi, Mauricio Reyes
Publisher	Springer Verlag
Pages	47-56
Number of pages	10
ISBN (Print)	9783030117221
DOIs	https://doi.org/10.1007/978-3-030-11723-8_5
State	Published - 2019
Event	4th International MICCAI Brainlesion Workshop, BrainLes 2018 held in conjunction with the Medical Image Computing for Computer Assisted Intervention Conference, MICCAI 2018 - Granada, Spain Duration: Sep 16 2018 → Sep 20 2018

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	11383 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	4th International MICCAI Brainlesion Workshop, BrainLes 2018 held in conjunction with the Medical Image Computing for Computer Assisted Intervention Conference, MICCAI 2018
Country/Territory	Spain
City	Granada
Period	9/16/18 → 9/20/18

Keywords

Inter-modal coupling
Logistic regression
Multiple sclerosis

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-030-11723-8_5

Cite this

Valcarcel, A. M., Linn, K. A., Khalid, F., Vandekar, S. N., Tauhid, S., Satterthwaite, T. D., Muschelli, J., Bakshi, R., & Shinohara, R. T. (2019). MIMoSA: An Approach to automatically segment T2 hyperintense and T1 hypointense lesions in multiple sclerosis. In F. Keyvan, S. Bakas, T. van Walsum, H. Kuijf, A. Crimi, & M. Reyes (Eds.), Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 4th International Workshop, BrainLes 2018, Held in Conjunction with MICCAI 2018, Revised Selected Papers (pp. 47-56). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11383 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-030-11723-8_5

MIMoSA: An Approach to automatically segment T2 hyperintense and T1 hypointense lesions in multiple sclerosis. / Valcarcel, Alessandra M.; Linn, Kristin A.; Khalid, Fariha et al.
Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 4th International Workshop, BrainLes 2018, Held in Conjunction with MICCAI 2018, Revised Selected Papers. ed. / Farahani Keyvan; Spyridon Bakas; Theo van Walsum; Hugo Kuijf; Alessandro Crimi; Mauricio Reyes. Springer Verlag, 2019. p. 47-56 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11383 LNCS).

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

Valcarcel, AM, Linn, KA, Khalid, F, Vandekar, SN, Tauhid, S, Satterthwaite, TD, Muschelli, J, Bakshi, R & Shinohara, RT 2019, MIMoSA: An Approach to automatically segment T2 hyperintense and T1 hypointense lesions in multiple sclerosis. in F Keyvan, S Bakas, T van Walsum, H Kuijf, A Crimi & M Reyes (eds), Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 4th International Workshop, BrainLes 2018, Held in Conjunction with MICCAI 2018, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11383 LNCS, Springer Verlag, pp. 47-56, 4th International MICCAI Brainlesion Workshop, BrainLes 2018 held in conjunction with the Medical Image Computing for Computer Assisted Intervention Conference, MICCAI 2018, Granada, Spain, 9/16/18. https://doi.org/10.1007/978-3-030-11723-8_5

Valcarcel AM, Linn KA, Khalid F, Vandekar SN, Tauhid S, Satterthwaite TD et al. MIMoSA: An Approach to automatically segment T2 hyperintense and T1 hypointense lesions in multiple sclerosis. In Keyvan F, Bakas S, van Walsum T, Kuijf H, Crimi A, Reyes M, editors, Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 4th International Workshop, BrainLes 2018, Held in Conjunction with MICCAI 2018, Revised Selected Papers. Springer Verlag. 2019. p. 47-56. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-11723-8_5

Valcarcel, Alessandra M. ; Linn, Kristin A. ; Khalid, Fariha et al. / MIMoSA : An Approach to automatically segment T2 hyperintense and T1 hypointense lesions in multiple sclerosis. Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 4th International Workshop, BrainLes 2018, Held in Conjunction with MICCAI 2018, Revised Selected Papers. editor / Farahani Keyvan ; Spyridon Bakas ; Theo van Walsum ; Hugo Kuijf ; Alessandro Crimi ; Mauricio Reyes. Springer Verlag, 2019. pp. 47-56 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{a08f0392febd4b56ae7f8f2d2ce006b0,

title = "MIMoSA: An Approach to automatically segment T2 hyperintense and T1 hypointense lesions in multiple sclerosis",

abstract = "Magnetic resonance imaging (MRI) is crucial for in vivo detection and characterization of white matter lesions (WML) in multiple sclerosis (MS). The most widely established MRI outcome measure is the volume of hyperintense lesions on T2-weighted images (T2L). Unfortunately, T2L are non-specific for the level of tissue destruction and show a weak relationship to clinical status. Interest in lesions appearing hypointense on T1-weighted images (T1L) (“black holes”), which provide more specificity for axonal loss and a closer link to neurologic disability, has thus grown. The technical difficulty of T1L segmentation has led investigators to rely on time-consuming manual assessments prone to inter- and intra-rater variability. We implement MIMoSA, a current T2L automatic segmentation approach, to delineate T1L. Using cross-validation, MIMoSA proved robust for segmenting both T2L and T1L. For T2L, a S{\o}rensen-Dice coefficient (DSC) of 0.6 and partial AUC (pAUC) up to 1% false positive rate of 0.69 were achieved. For T1L, 0.48 DSC and 0.63 pAUC were achieved. The correlation between EDSS and manual versus automatic volumes were similar for T1L (0.32 manual vs. 0.34 MIMoSA) and T2L (0.34 vs. 0.34).",

keywords = "Inter-modal coupling, Logistic regression, Multiple sclerosis",

author = "Valcarcel, {Alessandra M.} and Linn, {Kristin A.} and Fariha Khalid and Vandekar, {Simon N.} and Shahamat Tauhid and Satterthwaite, {Theodore D.} and John Muschelli and Rohit Bakshi and Shinohara, {Russell T.}",

note = "Publisher Copyright: {\textcopyright} Springer Nature Switzerland AG 2019.; 4th International MICCAI Brainlesion Workshop, BrainLes 2018 held in conjunction with the Medical Image Computing for Computer Assisted Intervention Conference, MICCAI 2018 ; Conference date: 16-09-2018 Through 20-09-2018",

year = "2019",

doi = "10.1007/978-3-030-11723-8_5",

language = "English (US)",

isbn = "9783030117221",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "47--56",

editor = "Farahani Keyvan and Spyridon Bakas and {van Walsum}, Theo and Hugo Kuijf and Alessandro Crimi and Mauricio Reyes",

booktitle = "Brainlesion",

}

TY - GEN

T1 - MIMoSA

T2 - 4th International MICCAI Brainlesion Workshop, BrainLes 2018 held in conjunction with the Medical Image Computing for Computer Assisted Intervention Conference, MICCAI 2018

AU - Valcarcel, Alessandra M.

AU - Linn, Kristin A.

AU - Khalid, Fariha

AU - Vandekar, Simon N.

AU - Tauhid, Shahamat

AU - Satterthwaite, Theodore D.

AU - Muschelli, John

AU - Bakshi, Rohit

AU - Shinohara, Russell T.

N1 - Publisher Copyright: © Springer Nature Switzerland AG 2019.

PY - 2019

Y1 - 2019

N2 - Magnetic resonance imaging (MRI) is crucial for in vivo detection and characterization of white matter lesions (WML) in multiple sclerosis (MS). The most widely established MRI outcome measure is the volume of hyperintense lesions on T2-weighted images (T2L). Unfortunately, T2L are non-specific for the level of tissue destruction and show a weak relationship to clinical status. Interest in lesions appearing hypointense on T1-weighted images (T1L) (“black holes”), which provide more specificity for axonal loss and a closer link to neurologic disability, has thus grown. The technical difficulty of T1L segmentation has led investigators to rely on time-consuming manual assessments prone to inter- and intra-rater variability. We implement MIMoSA, a current T2L automatic segmentation approach, to delineate T1L. Using cross-validation, MIMoSA proved robust for segmenting both T2L and T1L. For T2L, a Sørensen-Dice coefficient (DSC) of 0.6 and partial AUC (pAUC) up to 1% false positive rate of 0.69 were achieved. For T1L, 0.48 DSC and 0.63 pAUC were achieved. The correlation between EDSS and manual versus automatic volumes were similar for T1L (0.32 manual vs. 0.34 MIMoSA) and T2L (0.34 vs. 0.34).

AB - Magnetic resonance imaging (MRI) is crucial for in vivo detection and characterization of white matter lesions (WML) in multiple sclerosis (MS). The most widely established MRI outcome measure is the volume of hyperintense lesions on T2-weighted images (T2L). Unfortunately, T2L are non-specific for the level of tissue destruction and show a weak relationship to clinical status. Interest in lesions appearing hypointense on T1-weighted images (T1L) (“black holes”), which provide more specificity for axonal loss and a closer link to neurologic disability, has thus grown. The technical difficulty of T1L segmentation has led investigators to rely on time-consuming manual assessments prone to inter- and intra-rater variability. We implement MIMoSA, a current T2L automatic segmentation approach, to delineate T1L. Using cross-validation, MIMoSA proved robust for segmenting both T2L and T1L. For T2L, a Sørensen-Dice coefficient (DSC) of 0.6 and partial AUC (pAUC) up to 1% false positive rate of 0.69 were achieved. For T1L, 0.48 DSC and 0.63 pAUC were achieved. The correlation between EDSS and manual versus automatic volumes were similar for T1L (0.32 manual vs. 0.34 MIMoSA) and T2L (0.34 vs. 0.34).

KW - Inter-modal coupling

KW - Logistic regression

KW - Multiple sclerosis

UR - http://www.scopus.com/inward/record.url?scp=85063427651&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85063427651&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-11723-8_5

DO - 10.1007/978-3-030-11723-8_5

M3 - Conference contribution

AN - SCOPUS:85063427651

SN - 9783030117221

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

SP - 47

EP - 56

BT - Brainlesion

A2 - Keyvan, Farahani

A2 - Bakas, Spyridon

A2 - van Walsum, Theo

A2 - Kuijf, Hugo

A2 - Crimi, Alessandro

A2 - Reyes, Mauricio

PB - Springer Verlag

Y2 - 16 September 2018 through 20 September 2018

ER -

MIMoSA: An Approach to automatically segment T2 hyperintense and T1 hypointense lesions in multiple sclerosis

Abstract

Publication series

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this