Gp-Unet: Lesion detection from weak labels with a 3D regression network

Florian Dubost; Gerda Bortsova; Hieab H. Adams; M. Arfan Ikram; Wiro J. Niessen; Meike W. Vernooij; Marleen de Bruijne

doi:10.1007/978-3-319-66179-7_25

Gp-Unet: Lesion detection from weak labels with a 3D regression network

Florian Dubost^*, Gerda Bortsova, Hieab H. Adams, M. Arfan Ikram, Wiro J. Niessen, Meike W. Vernooij, Marleen de Bruijne

^*Corresponding author for this work

ImPhys/Imaging Physics

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

44 Citations (Scopus)

Abstract

We propose a novel convolutional neural network for lesion detection from weak labels. Only a single, global label per image - the lesion count - is needed for training. We train a regression network with a fully convolutional architecture combined with a global pooling layer to aggregate the 3D output into a scalar indicating the lesion count. When testing on unseen images, we first run the network to estimate the number of lesions. Then we remove the global pooling layer to compute localization maps of the size of the input image. We evaluate the proposed network on the detection of enlarged perivascular spaces in the basal ganglia in MRI. Our method achieves a sensitivity of 62% with on average 1.5 false positives per image. Compared with four other approaches based on intensity thresholding, saliency and class maps, our method has a 20% higher sensitivity.

Original language	English
Title of host publication	Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings
Editors	M. Descoteaux
Publisher	Springer
Pages	214-221
Number of pages	8
Volume	10435 LNCS
ISBN (Print)	9783319661780
DOIs	https://doi.org/10.1007/978-3-319-66179-7_25
Publication status	Published - 2017
Event	Medical Image Computing and Computer-Assisted Intervention, MICCAI 2017: 20th International Conference - Quebec City, Canada Duration: 11 Sept 2017 → 13 Sept 2017

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	10435 LNCS
ISSN (Print)	03029743
ISSN (Electronic)	16113349

Conference

Conference	Medical Image Computing and Computer-Assisted Intervention, MICCAI 2017
Country/Territory	Canada
City	Quebec City
Period	11/09/17 → 13/09/17

Access to Document

10.1007/978-3-319-66179-7_25

Cite this

Dubost, F., Bortsova, G., Adams, H. H., Ikram, M. A., Niessen, W. J., Vernooij, M. W., & de Bruijne, M. (2017). Gp-Unet: Lesion detection from weak labels with a 3D regression network. In M. Descoteaux (Ed.), Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings (Vol. 10435 LNCS, pp. 214-221). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10435 LNCS). Springer. https://doi.org/10.1007/978-3-319-66179-7_25

Dubost, Florian ; Bortsova, Gerda ; Adams, Hieab H. et al. / Gp-Unet : Lesion detection from weak labels with a 3D regression network. Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings. editor / M. Descoteaux. Vol. 10435 LNCS Springer, 2017. pp. 214-221 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{76dc3101167b48ccbd92c350d0e65a62,

title = "Gp-Unet: Lesion detection from weak labels with a 3D regression network",

abstract = "We propose a novel convolutional neural network for lesion detection from weak labels. Only a single, global label per image - the lesion count - is needed for training. We train a regression network with a fully convolutional architecture combined with a global pooling layer to aggregate the 3D output into a scalar indicating the lesion count. When testing on unseen images, we first run the network to estimate the number of lesions. Then we remove the global pooling layer to compute localization maps of the size of the input image. We evaluate the proposed network on the detection of enlarged perivascular spaces in the basal ganglia in MRI. Our method achieves a sensitivity of 62% with on average 1.5 false positives per image. Compared with four other approaches based on intensity thresholding, saliency and class maps, our method has a 20% higher sensitivity.",

author = "Florian Dubost and Gerda Bortsova and Adams, {Hieab H.} and Ikram, {M. Arfan} and Niessen, {Wiro J.} and Vernooij, {Meike W.} and {de Bruijne}, Marleen",

year = "2017",

doi = "10.1007/978-3-319-66179-7_25",

language = "English",

isbn = "9783319661780",

volume = "10435 LNCS",

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

publisher = "Springer",

pages = "214--221",

editor = "M. Descoteaux",

booktitle = "Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings",

note = "Medical Image Computing and Computer-Assisted Intervention, MICCAI 2017 : 20th International Conference ; Conference date: 11-09-2017 Through 13-09-2017",

}

Dubost, F, Bortsova, G, Adams, HH, Ikram, MA, Niessen, WJ, Vernooij, MW & de Bruijne, M 2017, Gp-Unet: Lesion detection from weak labels with a 3D regression network. in M Descoteaux (ed.), Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings. vol. 10435 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10435 LNCS, Springer, pp. 214-221, Medical Image Computing and Computer-Assisted Intervention, MICCAI 2017, Quebec City, Canada, 11/09/17. https://doi.org/10.1007/978-3-319-66179-7_25

Gp-Unet: Lesion detection from weak labels with a 3D regression network. / Dubost, Florian; Bortsova, Gerda; Adams, Hieab H. et al.
Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings. ed. / M. Descoteaux. Vol. 10435 LNCS Springer, 2017. p. 214-221 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10435 LNCS).

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

TY - GEN

T1 - Gp-Unet

T2 - Medical Image Computing and Computer-Assisted Intervention, MICCAI 2017

AU - Dubost, Florian

AU - Bortsova, Gerda

AU - Adams, Hieab H.

AU - Ikram, M. Arfan

AU - Niessen, Wiro J.

AU - Vernooij, Meike W.

AU - de Bruijne, Marleen

PY - 2017

Y1 - 2017

N2 - We propose a novel convolutional neural network for lesion detection from weak labels. Only a single, global label per image - the lesion count - is needed for training. We train a regression network with a fully convolutional architecture combined with a global pooling layer to aggregate the 3D output into a scalar indicating the lesion count. When testing on unseen images, we first run the network to estimate the number of lesions. Then we remove the global pooling layer to compute localization maps of the size of the input image. We evaluate the proposed network on the detection of enlarged perivascular spaces in the basal ganglia in MRI. Our method achieves a sensitivity of 62% with on average 1.5 false positives per image. Compared with four other approaches based on intensity thresholding, saliency and class maps, our method has a 20% higher sensitivity.

AB - We propose a novel convolutional neural network for lesion detection from weak labels. Only a single, global label per image - the lesion count - is needed for training. We train a regression network with a fully convolutional architecture combined with a global pooling layer to aggregate the 3D output into a scalar indicating the lesion count. When testing on unseen images, we first run the network to estimate the number of lesions. Then we remove the global pooling layer to compute localization maps of the size of the input image. We evaluate the proposed network on the detection of enlarged perivascular spaces in the basal ganglia in MRI. Our method achieves a sensitivity of 62% with on average 1.5 false positives per image. Compared with four other approaches based on intensity thresholding, saliency and class maps, our method has a 20% higher sensitivity.

UR - http://www.scopus.com/inward/record.url?scp=85029548373&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-66179-7_25

DO - 10.1007/978-3-319-66179-7_25

M3 - Conference contribution

AN - SCOPUS:85029548373

SN - 9783319661780

VL - 10435 LNCS

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

SP - 214

EP - 221

BT - Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings

A2 - Descoteaux, M.

PB - Springer

Y2 - 11 September 2017 through 13 September 2017

ER -

Dubost F, Bortsova G, Adams HH, Ikram MA, Niessen WJ, Vernooij MW et al. Gp-Unet: Lesion detection from weak labels with a 3D regression network. In Descoteaux M, editor, Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings. Vol. 10435 LNCS. Springer. 2017. p. 214-221. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-66179-7_25

Gp-Unet: Lesion detection from weak labels with a 3D regression network

Abstract

Publication series

Conference

Access to Document

Other files and links

Fingerprint

Cite this