DC 2 U-Net: Tract Segmentation in Brain White Matter Using Dense Criss-Cross U-Net

Haoran Yin; Pengbo Xu; Hui Cui; Geng Chen; Jiquan Ma

doi:10.1007/978-3-031-21206-2_10

DC ² U-Net: Tract Segmentation in Brain White Matter Using Dense Criss-Cross U-Net

Haoran Yin
, Pengbo Xu
, Hui Cui
, Geng Chen
, Jiquan Ma

School of Computer Science

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

1 Scopus citations

Abstract

Diffusion magnetic resonance imaging (dMRI) is a non-invasive technique for studying the microstructure properties of brain white matter (WM) in vivo. Segmentation of WM fiber tracts can be used to characterize the topological structure of the human brain and to exploit the biological landmark of abnormal areas by dMRI. To improve the performance of the fiber tract segmentation, we propose a novel U-Net based architecture with dense criss-cross attention, which captures non-local rich global contextual information more efficiently. Our model is evaluated using the real brain data from Human Connectome Project (HCP). Extensive experiments demonstrate that our model improves the performance of fiber tract segmentation, especially for the fiber bundle with complicated topology structure.

Original language	English
Title of host publication	Computational Diffusion MRI - 13th International Workshop, CDMRI 2022, Held in Conjunction with MICCAI 2022, Proceedings
Editors	Suheyla Cetin-Karayumak, Daan Christiaens, Matteo Figini, Pamela Guevara, Tomasz Pieciak, Elizabeth Powell, Francois Rheault
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	115-124
Number of pages	10
ISBN (Print)	9783031212055
DOIs	https://doi.org/10.1007/978-3-031-21206-2_10
State	Published - 2022
Event	13th International Workshop on Computational Diffusion MRI, CDMRI 2022 Held in Conjunction with MICCAI 2022 - Singapore, Singapore Duration: 22 Sep 2022 → 22 Sep 2022

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	13722 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	13th International Workshop on Computational Diffusion MRI, CDMRI 2022 Held in Conjunction with MICCAI 2022
Country/Territory	Singapore
City	Singapore
Period	22/09/22 → 22/09/22

Keywords

Attention
Dense connection
Diffusion MRI
Fiber tract segmentation

Access to Document

10.1007/978-3-031-21206-2_10

Cite this

Yin, H., Xu, P., Cui, H., Chen, G., & Ma, J. (2022). DC ² U-Net: Tract Segmentation in Brain White Matter Using Dense Criss-Cross U-Net. In S. Cetin-Karayumak, D. Christiaens, M. Figini, P. Guevara, T. Pieciak, E. Powell, & F. Rheault (Eds.), Computational Diffusion MRI - 13th International Workshop, CDMRI 2022, Held in Conjunction with MICCAI 2022, Proceedings (pp. 115-124). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13722 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-21206-2_10

Yin, Haoran ; Xu, Pengbo ; Cui, Hui et al. / DC ² U-Net : Tract Segmentation in Brain White Matter Using Dense Criss-Cross U-Net. Computational Diffusion MRI - 13th International Workshop, CDMRI 2022, Held in Conjunction with MICCAI 2022, Proceedings. editor / Suheyla Cetin-Karayumak ; Daan Christiaens ; Matteo Figini ; Pamela Guevara ; Tomasz Pieciak ; Elizabeth Powell ; Francois Rheault. Springer Science and Business Media Deutschland GmbH, 2022. pp. 115-124 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{3916f87f0eeb4986bcef3d9d8f6571da,

title = "DC 2 U-Net: Tract Segmentation in Brain White Matter Using Dense Criss-Cross U-Net",

abstract = "Diffusion magnetic resonance imaging (dMRI) is a non-invasive technique for studying the microstructure properties of brain white matter (WM) in vivo. Segmentation of WM fiber tracts can be used to characterize the topological structure of the human brain and to exploit the biological landmark of abnormal areas by dMRI. To improve the performance of the fiber tract segmentation, we propose a novel U-Net based architecture with dense criss-cross attention, which captures non-local rich global contextual information more efficiently. Our model is evaluated using the real brain data from Human Connectome Project (HCP). Extensive experiments demonstrate that our model improves the performance of fiber tract segmentation, especially for the fiber bundle with complicated topology structure.",

keywords = "Attention, Dense connection, Diffusion MRI, Fiber tract segmentation",

author = "Haoran Yin and Pengbo Xu and Hui Cui and Geng Chen and Jiquan Ma",

note = "Publisher Copyright: {\textcopyright} 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.; 13th International Workshop on Computational Diffusion MRI, CDMRI 2022 Held in Conjunction with MICCAI 2022 ; Conference date: 22-09-2022 Through 22-09-2022",

year = "2022",

doi = "10.1007/978-3-031-21206-2\_10",

language = "英语",

isbn = "9783031212055",

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

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "115--124",

editor = "Suheyla Cetin-Karayumak and Daan Christiaens and Matteo Figini and Pamela Guevara and Tomasz Pieciak and Elizabeth Powell and Francois Rheault",

booktitle = "Computational Diffusion MRI - 13th International Workshop, CDMRI 2022, Held in Conjunction with MICCAI 2022, Proceedings",

}

Yin, H, Xu, P, Cui, H, Chen, G & Ma, J 2022, DC ² U-Net: Tract Segmentation in Brain White Matter Using Dense Criss-Cross U-Net. in S Cetin-Karayumak, D Christiaens, M Figini, P Guevara, T Pieciak, E Powell & F Rheault (eds), Computational Diffusion MRI - 13th International Workshop, CDMRI 2022, Held in Conjunction with MICCAI 2022, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 13722 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 115-124, 13th International Workshop on Computational Diffusion MRI, CDMRI 2022 Held in Conjunction with MICCAI 2022, Singapore, Singapore, 22/09/22. https://doi.org/10.1007/978-3-031-21206-2_10

DC ² U-Net: Tract Segmentation in Brain White Matter Using Dense Criss-Cross U-Net. / Yin, Haoran; Xu, Pengbo; Cui, Hui et al.
Computational Diffusion MRI - 13th International Workshop, CDMRI 2022, Held in Conjunction with MICCAI 2022, Proceedings. ed. / Suheyla Cetin-Karayumak; Daan Christiaens; Matteo Figini; Pamela Guevara; Tomasz Pieciak; Elizabeth Powell; Francois Rheault. Springer Science and Business Media Deutschland GmbH, 2022. p. 115-124 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13722 LNCS).

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

TY - GEN

T1 - DC 2 U-Net

T2 - 13th International Workshop on Computational Diffusion MRI, CDMRI 2022 Held in Conjunction with MICCAI 2022

AU - Yin, Haoran

AU - Xu, Pengbo

AU - Cui, Hui

AU - Chen, Geng

AU - Ma, Jiquan

PY - 2022

Y1 - 2022

N2 - Diffusion magnetic resonance imaging (dMRI) is a non-invasive technique for studying the microstructure properties of brain white matter (WM) in vivo. Segmentation of WM fiber tracts can be used to characterize the topological structure of the human brain and to exploit the biological landmark of abnormal areas by dMRI. To improve the performance of the fiber tract segmentation, we propose a novel U-Net based architecture with dense criss-cross attention, which captures non-local rich global contextual information more efficiently. Our model is evaluated using the real brain data from Human Connectome Project (HCP). Extensive experiments demonstrate that our model improves the performance of fiber tract segmentation, especially for the fiber bundle with complicated topology structure.

AB - Diffusion magnetic resonance imaging (dMRI) is a non-invasive technique for studying the microstructure properties of brain white matter (WM) in vivo. Segmentation of WM fiber tracts can be used to characterize the topological structure of the human brain and to exploit the biological landmark of abnormal areas by dMRI. To improve the performance of the fiber tract segmentation, we propose a novel U-Net based architecture with dense criss-cross attention, which captures non-local rich global contextual information more efficiently. Our model is evaluated using the real brain data from Human Connectome Project (HCP). Extensive experiments demonstrate that our model improves the performance of fiber tract segmentation, especially for the fiber bundle with complicated topology structure.

KW - Attention

KW - Dense connection

KW - Diffusion MRI

KW - Fiber tract segmentation

UR - https://www.scopus.com/pages/publications/85144266150

U2 - 10.1007/978-3-031-21206-2_10

DO - 10.1007/978-3-031-21206-2_10

M3 - 会议稿件

AN - SCOPUS:85144266150

SN - 9783031212055

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

SP - 115

EP - 124

BT - Computational Diffusion MRI - 13th International Workshop, CDMRI 2022, Held in Conjunction with MICCAI 2022, Proceedings

A2 - Cetin-Karayumak, Suheyla

A2 - Christiaens, Daan

A2 - Figini, Matteo

A2 - Guevara, Pamela

A2 - Pieciak, Tomasz

A2 - Powell, Elizabeth

A2 - Rheault, Francois

PB - Springer Science and Business Media Deutschland GmbH

Y2 - 22 September 2022 through 22 September 2022

ER -

Yin H, Xu P, Cui H, Chen G, Ma J. DC ² U-Net: Tract Segmentation in Brain White Matter Using Dense Criss-Cross U-Net. In Cetin-Karayumak S, Christiaens D, Figini M, Guevara P, Pieciak T, Powell E, Rheault F, editors, Computational Diffusion MRI - 13th International Workshop, CDMRI 2022, Held in Conjunction with MICCAI 2022, Proceedings. Springer Science and Business Media Deutschland GmbH. 2022. p. 115-124. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-21206-2_10