Dice overlap measures for objects of unknown number: Application to lesion segmentation

Ipek Oguz; Aaron Carass; Dzung Pham; Snehashis Roy; Nagesh Subbana; Peter A. Calabresi; Paul A. Yushkevich; Russell T. Shinohara; Jerry L. Prince

doi:10.1007/978-3-319-75238-9_1

Dice overlap measures for objects of unknown number: Application to lesion segmentation

Ipek Oguz, Aaron Carass, Dzung Pham, Snehashis Roy, Nagesh Subbana, Peter A. Calabresi, Paul A. Yushkevich, Russell T. Shinohara, Jerry L. Prince

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

1 Scopus citations

Abstract

The Dice overlap ratio is commonly used to evaluate the performance of image segmentation algorithms. While Dice overlap is very useful as a standardized quantitative measure of segmentation accuracy in many applications, it offers a very limited picture of segmentation quality in complex segmentation tasks where the number of target objects is not known a priori, such as the segmentation of white matter lesions or lung nodules. While Dice overlap can still be used in these applications, segmentation algorithms may perform quite differently in ways not reflected by differences in their Dice score. Here we propose a new set of evaluation techniques that offer new insights into the behavior of segmentation algorithms. We illustrate these techniques with a case study comparing two popular multiple sclerosis (MS) lesion segmentation algorithms: OASIS and LesionTOADS.

Original language	English (US)
Title of host publication	Brainlesion
Subtitle of host publication	Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 3rd International Workshop, BrainLes 2017, Held in Conjunction with MICCAI 2017, Revised Selected Papers
Editors	Bjoern Menze, Alessandro Crimi, Hugo Kuijf, Mauricio Reyes, Spyridon Bakas
Publisher	Springer Verlag
Pages	3-14
Number of pages	12
ISBN (Print)	9783319752372
DOIs	https://doi.org/10.1007/978-3-319-75238-9_1
State	Published - 2018
Event	3rd International Workshop on Brainlesion, BrainLes 2017 Held in Conjunction with Medical Image Computing for Computer Assisted Intervention , MICCAI 2017 - Quebec City, Canada Duration: Sep 14 2017 → Sep 14 2017

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	10670 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Other

Other	3rd International Workshop on Brainlesion, BrainLes 2017 Held in Conjunction with Medical Image Computing for Computer Assisted Intervention , MICCAI 2017
Country/Territory	Canada
City	Quebec City
Period	9/14/17 → 9/14/17

Keywords

Evaluation
Lesion
MS
Segmentation

ASJC Scopus subject areas

Theoretical Computer Science
Computer Science(all)

Access to Document

10.1007/978-3-319-75238-9_1

Cite this

Oguz, I., Carass, A., Pham, D., Roy, S., Subbana, N., Calabresi, P. A., Yushkevich, P. A., Shinohara, R. T., & Prince, J. L. (2018). Dice overlap measures for objects of unknown number: Application to lesion segmentation. In B. Menze, A. Crimi, H. Kuijf, M. Reyes, & S. Bakas (Eds.), Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 3rd International Workshop, BrainLes 2017, Held in Conjunction with MICCAI 2017, Revised Selected Papers (pp. 3-14). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10670 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-75238-9_1

Dice overlap measures for objects of unknown number: Application to lesion segmentation. / Oguz, Ipek; Carass, Aaron; Pham, Dzung et al.
Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 3rd International Workshop, BrainLes 2017, Held in Conjunction with MICCAI 2017, Revised Selected Papers. ed. / Bjoern Menze; Alessandro Crimi; Hugo Kuijf; Mauricio Reyes; Spyridon Bakas. Springer Verlag, 2018. p. 3-14 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10670 LNCS).

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

Oguz, I, Carass, A, Pham, D, Roy, S, Subbana, N, Calabresi, PA, Yushkevich, PA, Shinohara, RT & Prince, JL 2018, Dice overlap measures for objects of unknown number: Application to lesion segmentation. in B Menze, A Crimi, H Kuijf, M Reyes & S Bakas (eds), Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 3rd International Workshop, BrainLes 2017, Held in Conjunction with MICCAI 2017, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10670 LNCS, Springer Verlag, pp. 3-14, 3rd International Workshop on Brainlesion, BrainLes 2017 Held in Conjunction with Medical Image Computing for Computer Assisted Intervention , MICCAI 2017, Quebec City, Canada, 9/14/17. https://doi.org/10.1007/978-3-319-75238-9_1

Oguz I, Carass A, Pham D, Roy S, Subbana N, Calabresi PA et al. Dice overlap measures for objects of unknown number: Application to lesion segmentation. In Menze B, Crimi A, Kuijf H, Reyes M, Bakas S, editors, Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 3rd International Workshop, BrainLes 2017, Held in Conjunction with MICCAI 2017, Revised Selected Papers. Springer Verlag. 2018. p. 3-14. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-75238-9_1

Oguz, Ipek ; Carass, Aaron ; Pham, Dzung et al. / Dice overlap measures for objects of unknown number : Application to lesion segmentation. Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 3rd International Workshop, BrainLes 2017, Held in Conjunction with MICCAI 2017, Revised Selected Papers. editor / Bjoern Menze ; Alessandro Crimi ; Hugo Kuijf ; Mauricio Reyes ; Spyridon Bakas. Springer Verlag, 2018. pp. 3-14 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{c131f37fed2e4bf28c1f3f590fcfbd44,

title = "Dice overlap measures for objects of unknown number: Application to lesion segmentation",

abstract = "The Dice overlap ratio is commonly used to evaluate the performance of image segmentation algorithms. While Dice overlap is very useful as a standardized quantitative measure of segmentation accuracy in many applications, it offers a very limited picture of segmentation quality in complex segmentation tasks where the number of target objects is not known a priori, such as the segmentation of white matter lesions or lung nodules. While Dice overlap can still be used in these applications, segmentation algorithms may perform quite differently in ways not reflected by differences in their Dice score. Here we propose a new set of evaluation techniques that offer new insights into the behavior of segmentation algorithms. We illustrate these techniques with a case study comparing two popular multiple sclerosis (MS) lesion segmentation algorithms: OASIS and LesionTOADS.",

keywords = "Evaluation, Lesion, MS, Segmentation",

author = "Ipek Oguz and Aaron Carass and Dzung Pham and Snehashis Roy and Nagesh Subbana and Calabresi, {Peter A.} and Yushkevich, {Paul A.} and Shinohara, {Russell T.} and Prince, {Jerry L.}",

note = "Funding Information: Acknowledgments. This work was supported, in part, by NIH grants NINDS R01-NS094456, NINDS R01-NS085211, NINDS R21-NS093349, NIBIB R01-EB017255, NINDS R01-NS082347, NINDS R01-NS070906, as well as National MS Society grant RG-1507-05243. Publisher Copyright: {\textcopyright} Springer International Publishing AG, part of Springer Nature 2018.; 3rd International Workshop on Brainlesion, BrainLes 2017 Held in Conjunction with Medical Image Computing for Computer Assisted Intervention , MICCAI 2017 ; Conference date: 14-09-2017 Through 14-09-2017",

year = "2018",

doi = "10.1007/978-3-319-75238-9_1",

language = "English (US)",

isbn = "9783319752372",

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

publisher = "Springer Verlag",

pages = "3--14",

editor = "Bjoern Menze and Alessandro Crimi and Hugo Kuijf and Mauricio Reyes and Spyridon Bakas",

booktitle = "Brainlesion",

}

TY - GEN

T1 - Dice overlap measures for objects of unknown number

T2 - 3rd International Workshop on Brainlesion, BrainLes 2017 Held in Conjunction with Medical Image Computing for Computer Assisted Intervention , MICCAI 2017

AU - Oguz, Ipek

AU - Carass, Aaron

AU - Pham, Dzung

AU - Roy, Snehashis

AU - Subbana, Nagesh

AU - Calabresi, Peter A.

AU - Yushkevich, Paul A.

AU - Shinohara, Russell T.

AU - Prince, Jerry L.

N1 - Funding Information: Acknowledgments. This work was supported, in part, by NIH grants NINDS R01-NS094456, NINDS R01-NS085211, NINDS R21-NS093349, NIBIB R01-EB017255, NINDS R01-NS082347, NINDS R01-NS070906, as well as National MS Society grant RG-1507-05243. Publisher Copyright: © Springer International Publishing AG, part of Springer Nature 2018.

PY - 2018

Y1 - 2018

N2 - The Dice overlap ratio is commonly used to evaluate the performance of image segmentation algorithms. While Dice overlap is very useful as a standardized quantitative measure of segmentation accuracy in many applications, it offers a very limited picture of segmentation quality in complex segmentation tasks where the number of target objects is not known a priori, such as the segmentation of white matter lesions or lung nodules. While Dice overlap can still be used in these applications, segmentation algorithms may perform quite differently in ways not reflected by differences in their Dice score. Here we propose a new set of evaluation techniques that offer new insights into the behavior of segmentation algorithms. We illustrate these techniques with a case study comparing two popular multiple sclerosis (MS) lesion segmentation algorithms: OASIS and LesionTOADS.

AB - The Dice overlap ratio is commonly used to evaluate the performance of image segmentation algorithms. While Dice overlap is very useful as a standardized quantitative measure of segmentation accuracy in many applications, it offers a very limited picture of segmentation quality in complex segmentation tasks where the number of target objects is not known a priori, such as the segmentation of white matter lesions or lung nodules. While Dice overlap can still be used in these applications, segmentation algorithms may perform quite differently in ways not reflected by differences in their Dice score. Here we propose a new set of evaluation techniques that offer new insights into the behavior of segmentation algorithms. We illustrate these techniques with a case study comparing two popular multiple sclerosis (MS) lesion segmentation algorithms: OASIS and LesionTOADS.

KW - Evaluation

KW - Lesion

KW - MS

KW - Segmentation

UR - http://www.scopus.com/inward/record.url?scp=85042530721&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85042530721&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-75238-9_1

DO - 10.1007/978-3-319-75238-9_1

M3 - Conference contribution

C2 - 29714358

AN - SCOPUS:85042530721

SN - 9783319752372

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

SP - 3

EP - 14

BT - Brainlesion

A2 - Menze, Bjoern

A2 - Crimi, Alessandro

A2 - Kuijf, Hugo

A2 - Reyes, Mauricio

A2 - Bakas, Spyridon

PB - Springer Verlag

Y2 - 14 September 2017 through 14 September 2017

ER -

Dice overlap measures for objects of unknown number: Application to lesion segmentation

Abstract

Publication series

Other

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this