Does manual delineation only provide the side information in CT prostate segmentation?

Yinghuan Shi, Wanqi Yang, Yang Gao, Dinggang Shen

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

5 Citations (Scopus)

Abstract

Prostate segmentation, for accurate prostate localization in CT images, is regarded as a crucial yet challenging task. Nevertheless, due to the inevitable factors (e.g., low contrast, large appearance and shape changes), the most important problem is how to learn the informative feature representation to distinguish the prostate from non-prostate regions. We address this challenging feature learning by leveraging the manual delineation as guidance: the manual delineation does not only indicate the category of patches, but also helps enhance the appearance of prostate. This is realized by the proposed cascaded deep domain adaptation (CDDA) model. Specifically, CDDA constructs several consecutive source domains by employing a mask of manual delineation to overlay on the original CT images with different mask ratios. Upon these source domains, convnet will guide better transferrable feature learning until to the target domain. Particularly, we implement two typical methods: patch-to-scalar (CDDA-CNN) and patch-to-patch (CDDA-FCN). Also, we theoretically analyze the generalization error bound of CDDA. Experimental results show the promising results of our method.

Original languageEnglish
Title of host publicationMedical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings
PublisherSpringer Verlag
Pages692-700
Number of pages9
Volume10435 LNCS
ISBN (Print)9783319661780
DOIs
Publication statusPublished - 2017 Jan 1
Externally publishedYes
Event20th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2017 - Quebec City, Canada
Duration: 2017 Sep 112017 Sep 13

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume10435 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other20th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2017
CountryCanada
CityQuebec City
Period17/9/1117/9/13

Fingerprint

Side Information
Masks
Segmentation
Patch
CT Image
Mask
Generalization Error
Overlay
Error Bounds
Guidance
Consecutive
Scalar
Target
Experimental Results

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Shi, Y., Yang, W., Gao, Y., & Shen, D. (2017). Does manual delineation only provide the side information in CT prostate segmentation? In Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings (Vol. 10435 LNCS, pp. 692-700). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10435 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-66179-7_79

Does manual delineation only provide the side information in CT prostate segmentation? / Shi, Yinghuan; Yang, Wanqi; Gao, Yang; Shen, Dinggang.

Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings. Vol. 10435 LNCS Springer Verlag, 2017. p. 692-700 (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/Report/Conference proceedingConference contribution

Shi, Y, Yang, W, Gao, Y & Shen, D 2017, Does manual delineation only provide the side information in CT prostate segmentation? in 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 Verlag, pp. 692-700, 20th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2017, Quebec City, Canada, 17/9/11. https://doi.org/10.1007/978-3-319-66179-7_79
Shi Y, Yang W, Gao Y, Shen D. Does manual delineation only provide the side information in CT prostate segmentation? In Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings. Vol. 10435 LNCS. Springer Verlag. 2017. p. 692-700. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-66179-7_79
Shi, Yinghuan ; Yang, Wanqi ; Gao, Yang ; Shen, Dinggang. / Does manual delineation only provide the side information in CT prostate segmentation?. Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings. Vol. 10435 LNCS Springer Verlag, 2017. pp. 692-700 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{70e6308fd99442999da5afcbac7361a2,
title = "Does manual delineation only provide the side information in CT prostate segmentation?",
abstract = "Prostate segmentation, for accurate prostate localization in CT images, is regarded as a crucial yet challenging task. Nevertheless, due to the inevitable factors (e.g., low contrast, large appearance and shape changes), the most important problem is how to learn the informative feature representation to distinguish the prostate from non-prostate regions. We address this challenging feature learning by leveraging the manual delineation as guidance: the manual delineation does not only indicate the category of patches, but also helps enhance the appearance of prostate. This is realized by the proposed cascaded deep domain adaptation (CDDA) model. Specifically, CDDA constructs several consecutive source domains by employing a mask of manual delineation to overlay on the original CT images with different mask ratios. Upon these source domains, convnet will guide better transferrable feature learning until to the target domain. Particularly, we implement two typical methods: patch-to-scalar (CDDA-CNN) and patch-to-patch (CDDA-FCN). Also, we theoretically analyze the generalization error bound of CDDA. Experimental results show the promising results of our method.",
author = "Yinghuan Shi and Wanqi Yang and Yang Gao and Dinggang Shen",
year = "2017",
month = "1",
day = "1",
doi = "10.1007/978-3-319-66179-7_79",
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 Verlag",
pages = "692--700",
booktitle = "Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings",

}

TY - GEN

T1 - Does manual delineation only provide the side information in CT prostate segmentation?

AU - Shi, Yinghuan

AU - Yang, Wanqi

AU - Gao, Yang

AU - Shen, Dinggang

PY - 2017/1/1

Y1 - 2017/1/1

N2 - Prostate segmentation, for accurate prostate localization in CT images, is regarded as a crucial yet challenging task. Nevertheless, due to the inevitable factors (e.g., low contrast, large appearance and shape changes), the most important problem is how to learn the informative feature representation to distinguish the prostate from non-prostate regions. We address this challenging feature learning by leveraging the manual delineation as guidance: the manual delineation does not only indicate the category of patches, but also helps enhance the appearance of prostate. This is realized by the proposed cascaded deep domain adaptation (CDDA) model. Specifically, CDDA constructs several consecutive source domains by employing a mask of manual delineation to overlay on the original CT images with different mask ratios. Upon these source domains, convnet will guide better transferrable feature learning until to the target domain. Particularly, we implement two typical methods: patch-to-scalar (CDDA-CNN) and patch-to-patch (CDDA-FCN). Also, we theoretically analyze the generalization error bound of CDDA. Experimental results show the promising results of our method.

AB - Prostate segmentation, for accurate prostate localization in CT images, is regarded as a crucial yet challenging task. Nevertheless, due to the inevitable factors (e.g., low contrast, large appearance and shape changes), the most important problem is how to learn the informative feature representation to distinguish the prostate from non-prostate regions. We address this challenging feature learning by leveraging the manual delineation as guidance: the manual delineation does not only indicate the category of patches, but also helps enhance the appearance of prostate. This is realized by the proposed cascaded deep domain adaptation (CDDA) model. Specifically, CDDA constructs several consecutive source domains by employing a mask of manual delineation to overlay on the original CT images with different mask ratios. Upon these source domains, convnet will guide better transferrable feature learning until to the target domain. Particularly, we implement two typical methods: patch-to-scalar (CDDA-CNN) and patch-to-patch (CDDA-FCN). Also, we theoretically analyze the generalization error bound of CDDA. Experimental results show the promising results of our method.

UR - http://www.scopus.com/inward/record.url?scp=85029525475&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85029525475&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-66179-7_79

DO - 10.1007/978-3-319-66179-7_79

M3 - Conference contribution

AN - SCOPUS:85029525475

SN - 9783319661780

VL - 10435 LNCS

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

SP - 692

EP - 700

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

PB - Springer Verlag

ER -