Explicit Topological Priors for Deep-Learning Based Image Segmentation Using Persistent Homology

James R. Clough; Ilkay Oksuz; Nicholas Byrne; Julia A. Schnabel; Andrew P. King

doi:10.1007/978-3-030-20351-1_2

Explicit Topological Priors for Deep-Learning Based Image Segmentation Using Persistent Homology

James R. Clough^*
, Ilkay Oksuz
, Nicholas Byrne
, Julia A. Schnabel
, Andrew P. King

^*Bu çalışma için yazışmadan sorumlu yazar

King's College London

Araştırma sonucu: Kitap/Rapor/Konferans Bildirisinde Bölüm › Konferans katkısı › bilirkişi

49 Atıf (Scopus)

Özet

We present a novel method to explicitly incorporate topological prior knowledge into deep learning based segmentation, which is, to our knowledge, the first work to do so. Our method uses the concept of persistent homology, a tool from topological data analysis, to capture high-level topological characteristics of segmentation results in a way which is differentiable with respect to the pixelwise probability of being assigned to a given class. The topological prior knowledge consists of the sequence of desired Betti numbers of the segmentation. As a proof-of-concept we demonstrate our approach by applying it to the problem of left-ventricle segmentation of cardiac MR images of subjects from the UK Biobank dataset, where we show that it improves segmentation performance in terms of topological correctness without sacrificing pixelwise accuracy.

Orijinal dil	İngilizce
Ana bilgisayar yayını başlığı	Information Processing in Medical Imaging - 26th International Conference, IPMI 2019, Proceedings
Editörler	Siqi Bao, Albert C.S. Chung, James C. Gee, Paul A. Yushkevich
Yayınlayan	Springer Verlag
Sayfalar	16-28
Sayfa sayısı	13
ISBN (Basılı)	9783030203504
DOI'lar	https://doi.org/10.1007/978-3-030-20351-1_2
Yayın durumu	Yayınlandı - 2019
Harici olarak yayınlandı	Evet
Etkinlik	26th International Conference on Information Processing in Medical Imaging, IPMI 2019 - Hong Kong, China Süre: 2 Haz 2019 → 7 Haz 2019

Yayın serisi

Adı	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Hacim	11492 LNCS
ISSN (Basılı)	0302-9743
ISSN (Elektronik)	1611-3349

???event.eventtypes.event.conference???

???event.eventtypes.event.conference???	26th International Conference on Information Processing in Medical Imaging, IPMI 2019
Ülke/Bölge	China
Şehir	Hong Kong
Periyot	2/06/19 → 7/06/19

Bibliyografik not

Publisher Copyright:
© 2019, Springer Nature Switzerland AG.

Finansman

A.P. King—This work was supported by an EPSRC programme Grant (EP/P001009/1) and the Wellcome EPSRC Centre for Medical Engineering at School of Biomedical Engineering and Imaging Sciences, King’s College London (WT 203148/Z/16/Z). This research has been conducted using the UK Biobank Resource under Application Number 40119. We would like to thank Nvidia for kindly donating the Quadro P6000 GPU used in this research.

Finansörler	Finansör numarası
Wellcome EPSRC Centre for Medical Engineering at School of Biomedical Engineering and Imaging Sciences
King’s College London	WT 203148/Z/16/Z
Engineering and Physical Sciences Research Council	EP/P001009/1

Belgeye Erişim

10.1007/978-3-030-20351-1_2

Diğer Dosyalar ve Linkler

Link to publication in Scopus

Alıntı Yap

Clough, J. R., Oksuz, I., Byrne, N., Schnabel, J. A., & King, A. P. (2019). Explicit Topological Priors for Deep-Learning Based Image Segmentation Using Persistent Homology. In S. Bao, A. C. S. Chung, J. C. Gee, & P. A. Yushkevich (Eds.), Information Processing in Medical Imaging - 26th International Conference, IPMI 2019, Proceedings (pp. 16-28). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Hac. 11492 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-030-20351-1_2

Clough, James R. ; Oksuz, Ilkay ; Byrne, Nicholas et al. / Explicit Topological Priors for Deep-Learning Based Image Segmentation Using Persistent Homology. Information Processing in Medical Imaging - 26th International Conference, IPMI 2019, Proceedings. editör / Siqi Bao ; Albert C.S. Chung ; James C. Gee ; Paul A. Yushkevich. Springer Verlag, 2019. pp. 16-28 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{059884c5a1f34659a671aba4f3dd72b8,

title = "Explicit Topological Priors for Deep-Learning Based Image Segmentation Using Persistent Homology",

abstract = "We present a novel method to explicitly incorporate topological prior knowledge into deep learning based segmentation, which is, to our knowledge, the first work to do so. Our method uses the concept of persistent homology, a tool from topological data analysis, to capture high-level topological characteristics of segmentation results in a way which is differentiable with respect to the pixelwise probability of being assigned to a given class. The topological prior knowledge consists of the sequence of desired Betti numbers of the segmentation. As a proof-of-concept we demonstrate our approach by applying it to the problem of left-ventricle segmentation of cardiac MR images of subjects from the UK Biobank dataset, where we show that it improves segmentation performance in terms of topological correctness without sacrificing pixelwise accuracy.",

keywords = "Cardiac MRI, Persistent homology, Segmentation, Topological data analysis, Topology",

author = "Clough, \{James R.\} and Ilkay Oksuz and Nicholas Byrne and Schnabel, \{Julia A.\} and King, \{Andrew P.\}",

note = "Publisher Copyright: {\textcopyright} 2019, Springer Nature Switzerland AG.; 26th International Conference on Information Processing in Medical Imaging, IPMI 2019 ; Conference date: 02-06-2019 Through 07-06-2019",

year = "2019",

doi = "10.1007/978-3-030-20351-1\_2",

language = "English",

isbn = "9783030203504",

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

publisher = "Springer Verlag",

pages = "16--28",

editor = "Siqi Bao and Chung, \{Albert C.S.\} and Gee, \{James C.\} and Yushkevich, \{Paul A.\}",

booktitle = "Information Processing in Medical Imaging - 26th International Conference, IPMI 2019, Proceedings",

address = "Germany",

}

Clough, JR, Oksuz, I, Byrne, N, Schnabel, JA & King, AP 2019, Explicit Topological Priors for Deep-Learning Based Image Segmentation Using Persistent Homology. in S Bao, ACS Chung, JC Gee & PA Yushkevich (eds), Information Processing in Medical Imaging - 26th International Conference, IPMI 2019, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), hac. 11492 LNCS, Springer Verlag, pp. 16-28, 26th International Conference on Information Processing in Medical Imaging, IPMI 2019, Hong Kong, China, 2/06/19. https://doi.org/10.1007/978-3-030-20351-1_2

Explicit Topological Priors for Deep-Learning Based Image Segmentation Using Persistent Homology. / Clough, James R.; Oksuz, Ilkay; Byrne, Nicholas et al.
Information Processing in Medical Imaging - 26th International Conference, IPMI 2019, Proceedings. ed. / Siqi Bao; Albert C.S. Chung; James C. Gee; Paul A. Yushkevich. Springer Verlag, 2019. p. 16-28 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Hac. 11492 LNCS).

Araştırma sonucu: Kitap/Rapor/Konferans Bildirisinde Bölüm › Konferans katkısı › bilirkişi

TY - GEN

T1 - Explicit Topological Priors for Deep-Learning Based Image Segmentation Using Persistent Homology

AU - Clough, James R.

AU - Oksuz, Ilkay

AU - Byrne, Nicholas

AU - Schnabel, Julia A.

AU - King, Andrew P.

PY - 2019

Y1 - 2019

N2 - We present a novel method to explicitly incorporate topological prior knowledge into deep learning based segmentation, which is, to our knowledge, the first work to do so. Our method uses the concept of persistent homology, a tool from topological data analysis, to capture high-level topological characteristics of segmentation results in a way which is differentiable with respect to the pixelwise probability of being assigned to a given class. The topological prior knowledge consists of the sequence of desired Betti numbers of the segmentation. As a proof-of-concept we demonstrate our approach by applying it to the problem of left-ventricle segmentation of cardiac MR images of subjects from the UK Biobank dataset, where we show that it improves segmentation performance in terms of topological correctness without sacrificing pixelwise accuracy.

AB - We present a novel method to explicitly incorporate topological prior knowledge into deep learning based segmentation, which is, to our knowledge, the first work to do so. Our method uses the concept of persistent homology, a tool from topological data analysis, to capture high-level topological characteristics of segmentation results in a way which is differentiable with respect to the pixelwise probability of being assigned to a given class. The topological prior knowledge consists of the sequence of desired Betti numbers of the segmentation. As a proof-of-concept we demonstrate our approach by applying it to the problem of left-ventricle segmentation of cardiac MR images of subjects from the UK Biobank dataset, where we show that it improves segmentation performance in terms of topological correctness without sacrificing pixelwise accuracy.

KW - Cardiac MRI

KW - Persistent homology

KW - Segmentation

KW - Topological data analysis

KW - Topology

UR - https://www.scopus.com/pages/publications/85066130921

U2 - 10.1007/978-3-030-20351-1_2

DO - 10.1007/978-3-030-20351-1_2

M3 - Conference contribution

AN - SCOPUS:85066130921

SN - 9783030203504

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

SP - 16

EP - 28

BT - Information Processing in Medical Imaging - 26th International Conference, IPMI 2019, Proceedings

A2 - Bao, Siqi

A2 - Chung, Albert C.S.

A2 - Gee, James C.

A2 - Yushkevich, Paul A.

PB - Springer Verlag

T2 - 26th International Conference on Information Processing in Medical Imaging, IPMI 2019

Y2 - 2 June 2019 through 7 June 2019

ER -

Clough JR, Oksuz I, Byrne N, Schnabel JA, King AP. Explicit Topological Priors for Deep-Learning Based Image Segmentation Using Persistent Homology. In Bao S, Chung ACS, Gee JC, Yushkevich PA, editörler, Information Processing in Medical Imaging - 26th International Conference, IPMI 2019, Proceedings. Springer Verlag. 2019. p. 16-28. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-20351-1_2

Explicit Topological Priors for Deep-Learning Based Image Segmentation Using Persistent Homology

Özet

Yayın serisi

???event.eventtypes.event.conference???

Bibliyografik not

Finansman

Belgeye Erişim

Diğer Dosyalar ve Linkler

Parmak izi

Alıntı Yap