Learning features for tissue classification with the classification restricted Boltzmann machine

Gijs van Tulder; Marleen de Bruijne

doi:10.1007/978-3-319-13972-2_5

Learning features for tissue classification with the classification restricted Boltzmann machine

Gijs van Tulder^*, Marleen de Bruijne

^*Corresponding author af dette arbejde

Datalogisk Institut

15 Citationer (Scopus)

Abstract

Performance of automated tissue classification in medical imaging depends on the choice of descriptive features. In this paper, we show how restricted Boltzmann machines (RBMs) can be used to learn features that are especially suited for texture-based tissue classification. We introduce the convolutional classification RBM, a combination of the existing convolutional RBM and classification RBM, and use it for discriminative feature learning. We evaluate the classification accuracy of convolutional and non-convolutional classification RBMs on two lung CT problems. We find that RBM-learned features outperform conventional RBM-based feature learning, which is unsupervised and uses only a generative learning objective, as well as often-used filter banks. We show that a mixture of generative and discriminative learning can produce filters that give a higher classification accuracy.

Originalsprog	Engelsk
Titel	Medical Computer Vision: Algorithms for Big Data : International Workshop, MCV 2014, Held in Conjunction with MICCAI 2014, Cambridge, MA, USA, September 18, 2014, Revised Selected Papers
Redaktører	Bjoern Menze, Georg Langs, Albert Montillo, Michael Kelm, Henning Müller, Shaoting Zhang, Weidong (Tom) Cai, Dimitris Metaxas
Antal sider	12
Forlag	Springer
Publikationsdato	2014
Sider	47-58
Kapitel	5
ISBN (Trykt)	978-3-319-13971-5
ISBN (Elektronisk)	978-3-319-13972-2
DOI	https://doi.org/10.1007/978-3-319-13972-2_5
Status	Udgivet - 2014
Begivenhed	International Workshop on Medical Computer Vision 2014 - Cambridge, USA Varighed: 18 sep. 2014 → 18 sep. 2014

Konference

Konference	International Workshop on Medical Computer Vision 2014
Land/Område	USA
By	Cambridge
Periode	18/09/2014 → 18/09/2014

Navn	Lecture notes in computer science
ISSN	0302-9743

Adgang til dokumentet

10.1007/978-3-319-13972-2_5

Citationsformater

van Tulder, G., & de Bruijne, M. (2014). Learning features for tissue classification with the classification restricted Boltzmann machine. I B. Menze, G. Langs, A. Montillo, M. Kelm, H. Müller, S. Zhang, W. Cai, & D. Metaxas (red.), Medical Computer Vision: Algorithms for Big Data: International Workshop, MCV 2014, Held in Conjunction with MICCAI 2014, Cambridge, MA, USA, September 18, 2014, Revised Selected Papers (s. 47-58). Springer. Lecture notes in computer science https://doi.org/10.1007/978-3-319-13972-2_5

Learning features for tissue classification with the classification restricted Boltzmann machine. / van Tulder, Gijs; de Bruijne, Marleen.

Medical Computer Vision: Algorithms for Big Data: International Workshop, MCV 2014, Held in Conjunction with MICCAI 2014, Cambridge, MA, USA, September 18, 2014, Revised Selected Papers. red. / Bjoern Menze; Georg Langs; Albert Montillo; Michael Kelm; Henning Müller; Shaoting Zhang; Weidong (Tom) Cai; Dimitris Metaxas. Springer, 2014. s. 47-58 (Lecture notes in computer science).

Publikation: Bidrag til bog/antologi/rapport › Konferencebidrag i proceedings › Forskning › peer review

van Tulder, G & de Bruijne, M 2014, Learning features for tissue classification with the classification restricted Boltzmann machine. i B Menze, G Langs, A Montillo, M Kelm, H Müller, S Zhang, W Cai & D Metaxas (red), Medical Computer Vision: Algorithms for Big Data: International Workshop, MCV 2014, Held in Conjunction with MICCAI 2014, Cambridge, MA, USA, September 18, 2014, Revised Selected Papers. Springer, Lecture notes in computer science, s. 47-58, International Workshop on Medical Computer Vision 2014, Cambridge, USA, 18/09/2014. https://doi.org/10.1007/978-3-319-13972-2_5

van Tulder G, de Bruijne M. Learning features for tissue classification with the classification restricted Boltzmann machine. I Menze B, Langs G, Montillo A, Kelm M, Müller H, Zhang S, Cai W, Metaxas D, red., Medical Computer Vision: Algorithms for Big Data: International Workshop, MCV 2014, Held in Conjunction with MICCAI 2014, Cambridge, MA, USA, September 18, 2014, Revised Selected Papers. Springer. 2014. s. 47-58. (Lecture notes in computer science). doi: 10.1007/978-3-319-13972-2_5

van Tulder, Gijs ; de Bruijne, Marleen. / Learning features for tissue classification with the classification restricted Boltzmann machine. Medical Computer Vision: Algorithms for Big Data: International Workshop, MCV 2014, Held in Conjunction with MICCAI 2014, Cambridge, MA, USA, September 18, 2014, Revised Selected Papers. red. / Bjoern Menze ; Georg Langs ; Albert Montillo ; Michael Kelm ; Henning Müller ; Shaoting Zhang ; Weidong (Tom) Cai ; Dimitris Metaxas. Springer, 2014. s. 47-58 (Lecture notes in computer science).

@inproceedings{0042145b62b145fab22b90322f17078e,

title = "Learning features for tissue classification with the classification restricted Boltzmann machine",

abstract = "Performance of automated tissue classification in medical imaging depends on the choice of descriptive features. In this paper, we show how restricted Boltzmann machines (RBMs) can be used to learn features that are especially suited for texture-based tissue classification. We introduce the convolutional classification RBM, a combination of the existing convolutional RBM and classification RBM, and use it for discriminative feature learning. We evaluate the classification accuracy of convolutional and non-convolutional classification RBMs on two lung CT problems. We find that RBM-learned features outperform conventional RBM-based feature learning, which is unsupervised and uses only a generative learning objective, as well as often-used filter banks. We show that a mixture of generative and discriminative learning can produce filters that give a higher classification accuracy.",

author = "{van Tulder}, Gijs and {de Bruijne}, Marleen",

year = "2014",

doi = "10.1007/978-3-319-13972-2_5",

language = "English",

isbn = "978-3-319-13971-5",

series = "Lecture notes in computer science",

publisher = "Springer",

pages = "47--58",

editor = "Bjoern Menze and Georg Langs and Albert Montillo and Michael Kelm and Henning M{\"u}ller and Shaoting Zhang and Cai, {Weidong (Tom)} and Dimitris Metaxas",

booktitle = "Medical Computer Vision: Algorithms for Big Data",

note = "International Workshop on Medical Computer Vision 2014, MCV 2014 ; Conference date: 18-09-2014 Through 18-09-2014",

}

TY - GEN

T1 - Learning features for tissue classification with the classification restricted Boltzmann machine

AU - van Tulder, Gijs

AU - de Bruijne, Marleen

PY - 2014

Y1 - 2014

N2 - Performance of automated tissue classification in medical imaging depends on the choice of descriptive features. In this paper, we show how restricted Boltzmann machines (RBMs) can be used to learn features that are especially suited for texture-based tissue classification. We introduce the convolutional classification RBM, a combination of the existing convolutional RBM and classification RBM, and use it for discriminative feature learning. We evaluate the classification accuracy of convolutional and non-convolutional classification RBMs on two lung CT problems. We find that RBM-learned features outperform conventional RBM-based feature learning, which is unsupervised and uses only a generative learning objective, as well as often-used filter banks. We show that a mixture of generative and discriminative learning can produce filters that give a higher classification accuracy.

AB - Performance of automated tissue classification in medical imaging depends on the choice of descriptive features. In this paper, we show how restricted Boltzmann machines (RBMs) can be used to learn features that are especially suited for texture-based tissue classification. We introduce the convolutional classification RBM, a combination of the existing convolutional RBM and classification RBM, and use it for discriminative feature learning. We evaluate the classification accuracy of convolutional and non-convolutional classification RBMs on two lung CT problems. We find that RBM-learned features outperform conventional RBM-based feature learning, which is unsupervised and uses only a generative learning objective, as well as often-used filter banks. We show that a mixture of generative and discriminative learning can produce filters that give a higher classification accuracy.

U2 - 10.1007/978-3-319-13972-2_5

DO - 10.1007/978-3-319-13972-2_5

M3 - Article in proceedings

AN - SCOPUS:84917707434

SN - 978-3-319-13971-5

T3 - Lecture notes in computer science

SP - 47

EP - 58

BT - Medical Computer Vision: Algorithms for Big Data

A2 - Menze, Bjoern

A2 - Langs, Georg

A2 - Montillo, Albert

A2 - Kelm, Michael

A2 - Müller, Henning

A2 - Zhang, Shaoting

A2 - Cai, Weidong (Tom)

A2 - Metaxas, Dimitris

PB - Springer

T2 - International Workshop on Medical Computer Vision 2014

Y2 - 18 September 2014 through 18 September 2014

ER -

Learning features for tissue classification with the classification restricted Boltzmann machine

Abstract

Konference

Adgang til dokumentet

Fingeraftryk

Citationsformater