Longitudinal patch-based segmentation of multiple sclerosis white matter lesions

Snehashis Roy, Aaron Carass, Jerry Ladd Prince, Dzung L. Pham

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Segmenting T2-weighted white matter lesions from longitudinal MR images is essential in understanding progression of multiple sclerosis. Most lesion segmentation techniques find lesions independently at each time point, even though there are different noise and image contrast variations at each point in the time series. In this paper, we present a patch based 4D lesion segmentation method that takes advantage of the temporal component of longitudinal data. For each subject with multiple time-points, 4D patches are constructed from the T1-w and FLAIR scans of all time-points. For every 4D patch from a subject, a few relevant matching 4D patches are found from a reference, such that their convex combination reconstructs the subject’s 4D patch. Then corresponding manual segmentation patches of the reference are combined in a similar manner to generate a 4D membership of lesions of the subject patch. We compare our 4D patch-based segmentation with independent 3D voxel-based and patch-based lesion segmentation algorithms. Based on ground truth segmentations from 30 data sets, we show that the mean Dice coefficients between manual and automated segmentations improve after using the 4D approach compared to two state-of-the-art 3D segmentation algorithms.

Original languageEnglish (US)
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
PublisherSpringer Verlag
Pages194-202
Number of pages9
Volume9352
ISBN (Print)9783319248875
DOIs
StatePublished - 2015
Event6th International Workshop on Machine Learning in Medical Imaging, MLMI 2015 and Held in Conjunction with 18th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2015 - Munich, Germany
Duration: Oct 5 2015Oct 5 2015

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9352
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other6th International Workshop on Machine Learning in Medical Imaging, MLMI 2015 and Held in Conjunction with 18th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2015
CountryGermany
CityMunich
Period10/5/1510/5/15

Fingerprint

Multiple Sclerosis
Patch
Segmentation
Time series
Dice
Convex Combination
Voxel
Longitudinal Data
Progression

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Roy, S., Carass, A., Prince, J. L., & Pham, D. L. (2015). Longitudinal patch-based segmentation of multiple sclerosis white matter lesions. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 9352, pp. 194-202). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9352). Springer Verlag. https://doi.org/10.1007/978-3-319-24888-2_24

Longitudinal patch-based segmentation of multiple sclerosis white matter lesions. / Roy, Snehashis; Carass, Aaron; Prince, Jerry Ladd; Pham, Dzung L.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9352 Springer Verlag, 2015. p. 194-202 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9352).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Roy, S, Carass, A, Prince, JL & Pham, DL 2015, Longitudinal patch-based segmentation of multiple sclerosis white matter lesions. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 9352, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9352, Springer Verlag, pp. 194-202, 6th International Workshop on Machine Learning in Medical Imaging, MLMI 2015 and Held in Conjunction with 18th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2015, Munich, Germany, 10/5/15. https://doi.org/10.1007/978-3-319-24888-2_24
Roy S, Carass A, Prince JL, Pham DL. Longitudinal patch-based segmentation of multiple sclerosis white matter lesions. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9352. Springer Verlag. 2015. p. 194-202. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-24888-2_24
Roy, Snehashis ; Carass, Aaron ; Prince, Jerry Ladd ; Pham, Dzung L. / Longitudinal patch-based segmentation of multiple sclerosis white matter lesions. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9352 Springer Verlag, 2015. pp. 194-202 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{eab66c1cd0194be19605afa3038bfb7d,
title = "Longitudinal patch-based segmentation of multiple sclerosis white matter lesions",
abstract = "Segmenting T2-weighted white matter lesions from longitudinal MR images is essential in understanding progression of multiple sclerosis. Most lesion segmentation techniques find lesions independently at each time point, even though there are different noise and image contrast variations at each point in the time series. In this paper, we present a patch based 4D lesion segmentation method that takes advantage of the temporal component of longitudinal data. For each subject with multiple time-points, 4D patches are constructed from the T1-w and FLAIR scans of all time-points. For every 4D patch from a subject, a few relevant matching 4D patches are found from a reference, such that their convex combination reconstructs the subject’s 4D patch. Then corresponding manual segmentation patches of the reference are combined in a similar manner to generate a 4D membership of lesions of the subject patch. We compare our 4D patch-based segmentation with independent 3D voxel-based and patch-based lesion segmentation algorithms. Based on ground truth segmentations from 30 data sets, we show that the mean Dice coefficients between manual and automated segmentations improve after using the 4D approach compared to two state-of-the-art 3D segmentation algorithms.",
author = "Snehashis Roy and Aaron Carass and Prince, {Jerry Ladd} and Pham, {Dzung L.}",
year = "2015",
doi = "10.1007/978-3-319-24888-2_24",
language = "English (US)",
isbn = "9783319248875",
volume = "9352",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "194--202",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - Longitudinal patch-based segmentation of multiple sclerosis white matter lesions

AU - Roy, Snehashis

AU - Carass, Aaron

AU - Prince, Jerry Ladd

AU - Pham, Dzung L.

PY - 2015

Y1 - 2015

N2 - Segmenting T2-weighted white matter lesions from longitudinal MR images is essential in understanding progression of multiple sclerosis. Most lesion segmentation techniques find lesions independently at each time point, even though there are different noise and image contrast variations at each point in the time series. In this paper, we present a patch based 4D lesion segmentation method that takes advantage of the temporal component of longitudinal data. For each subject with multiple time-points, 4D patches are constructed from the T1-w and FLAIR scans of all time-points. For every 4D patch from a subject, a few relevant matching 4D patches are found from a reference, such that their convex combination reconstructs the subject’s 4D patch. Then corresponding manual segmentation patches of the reference are combined in a similar manner to generate a 4D membership of lesions of the subject patch. We compare our 4D patch-based segmentation with independent 3D voxel-based and patch-based lesion segmentation algorithms. Based on ground truth segmentations from 30 data sets, we show that the mean Dice coefficients between manual and automated segmentations improve after using the 4D approach compared to two state-of-the-art 3D segmentation algorithms.

AB - Segmenting T2-weighted white matter lesions from longitudinal MR images is essential in understanding progression of multiple sclerosis. Most lesion segmentation techniques find lesions independently at each time point, even though there are different noise and image contrast variations at each point in the time series. In this paper, we present a patch based 4D lesion segmentation method that takes advantage of the temporal component of longitudinal data. For each subject with multiple time-points, 4D patches are constructed from the T1-w and FLAIR scans of all time-points. For every 4D patch from a subject, a few relevant matching 4D patches are found from a reference, such that their convex combination reconstructs the subject’s 4D patch. Then corresponding manual segmentation patches of the reference are combined in a similar manner to generate a 4D membership of lesions of the subject patch. We compare our 4D patch-based segmentation with independent 3D voxel-based and patch-based lesion segmentation algorithms. Based on ground truth segmentations from 30 data sets, we show that the mean Dice coefficients between manual and automated segmentations improve after using the 4D approach compared to two state-of-the-art 3D segmentation algorithms.

UR - http://www.scopus.com/inward/record.url?scp=84952059686&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84952059686&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-24888-2_24

DO - 10.1007/978-3-319-24888-2_24

M3 - Conference contribution

C2 - 27570846

AN - SCOPUS:84952059686

SN - 9783319248875

VL - 9352

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

SP - 194

EP - 202

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

PB - Springer Verlag

ER -