Adaptive Semi-supervised Segmentation of Brain Vessels with Ambiguous Labels

Fengming Lin; Yan Xia; Nishant Ravikumar; Qiongyao Liu; Michael MacRaild; Alejandro F. Frangi

doi:10.48550/arXiv.2308.03613

Adaptive Semi-supervised Segmentation of Brain Vessels with Ambiguous Labels

Fengming Lin, Yan Xia, Nishant Ravikumar, Qiongyao Liu, Michael MacRaild, Alejandro F. Frangi^*

^*Corresponding author for this work

University of Leeds

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

Abstract

Accurate segmentation of brain vessels is crucial for cerebrovascular disease diagnosis and treatment. However, existing methods face challenges in capturing small vessels and handling datasets that are partially or ambiguously annotated. In this paper, we propose an adaptive semi-supervised approach to address these challenges. Our approach incorporates innovative techniques including progressive semi-supervised learning, adaptative training strategy, and boundary enhancement. Experimental results on 3DRA datasets demonstrate the superiority of our method in terms of mesh-based segmentation metrics. By leveraging the partially and ambiguously labeled data, which only annotates the main vessels, our method achieves impressive segmentation performance on mislabeled fine vessels, showcasing its potential for clinical applications.

Original language	English
Title of host publication	Data Augmentation, Labelling, and Imperfections - 3rd MICCAI Workshop, DALI 2023 Held in Conjunction with MICCAI 2023, Proceedings
Editors	Yuan Xue, Chen Chen, Chao Chen, Lianrui Zuo, Yihao Liu
Publisher	Springer Nature
Pages	106-116
Number of pages	11
ISBN (Print)	9783031581700
DOIs	https://doi.org/10.48550/arXiv.2308.03613 https://doi.org/10.1007/978-3-031-58171-7_11
Publication status	Published - 27 Apr 2024
Event	3rd International Workshop on Data Augmentation, Labeling, and Imperfections, DALI 2023 in conjunction with the 26th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2023 - Vancouver, Canada Duration: 12 Oct 2023 → 12 Oct 2023

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	14379 LNCS

Conference

Conference	3rd International Workshop on Data Augmentation, Labeling, and Imperfections, DALI 2023 in conjunction with the 26th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2023
Country/Territory	Canada
City	Vancouver
Period	12/10/23 → 12/10/23

Keywords

Adaptive model
Brain vessel segmentation
Semi-supervised learning

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

Cite this

Lin, F., Xia, Y., Ravikumar, N., Liu, Q., MacRaild, M., & Frangi, A. F. (2024). Adaptive Semi-supervised Segmentation of Brain Vessels with Ambiguous Labels. In Y. Xue, C. Chen, C. Chen, L. Zuo, & Y. Liu (Eds.), Data Augmentation, Labelling, and Imperfections - 3rd MICCAI Workshop, DALI 2023 Held in Conjunction with MICCAI 2023, Proceedings (pp. 106-116). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14379 LNCS). Springer Nature. https://doi.org/10.48550/arXiv.2308.03613, https://doi.org/10.1007/978-3-031-58171-7_11

Lin, Fengming ; Xia, Yan ; Ravikumar, Nishant et al. / Adaptive Semi-supervised Segmentation of Brain Vessels with Ambiguous Labels. Data Augmentation, Labelling, and Imperfections - 3rd MICCAI Workshop, DALI 2023 Held in Conjunction with MICCAI 2023, Proceedings. editor / Yuan Xue ; Chen Chen ; Chao Chen ; Lianrui Zuo ; Yihao Liu. Springer Nature, 2024. pp. 106-116 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{e723258f767044199f07e9234274f6ee,

title = "Adaptive Semi-supervised Segmentation of Brain Vessels with Ambiguous Labels",

abstract = "Accurate segmentation of brain vessels is crucial for cerebrovascular disease diagnosis and treatment. However, existing methods face challenges in capturing small vessels and handling datasets that are partially or ambiguously annotated. In this paper, we propose an adaptive semi-supervised approach to address these challenges. Our approach incorporates innovative techniques including progressive semi-supervised learning, adaptative training strategy, and boundary enhancement. Experimental results on 3DRA datasets demonstrate the superiority of our method in terms of mesh-based segmentation metrics. By leveraging the partially and ambiguously labeled data, which only annotates the main vessels, our method achieves impressive segmentation performance on mislabeled fine vessels, showcasing its potential for clinical applications.",

keywords = "Adaptive model, Brain vessel segmentation, Semi-supervised learning",

author = "Fengming Lin and Yan Xia and Nishant Ravikumar and Qiongyao Liu and Michael MacRaild and Frangi, {Alejandro F.}",

note = "Publisher Copyright: {\textcopyright} The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.; 3rd International Workshop on Data Augmentation, Labeling, and Imperfections, DALI 2023 in conjunction with the 26th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2023 ; Conference date: 12-10-2023 Through 12-10-2023",

year = "2024",

month = apr,

day = "27",

doi = "10.48550/arXiv.2308.03613",

language = "English",

isbn = "9783031581700",

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

publisher = "Springer Nature",

pages = "106--116",

editor = "Yuan Xue and Chen Chen and Chao Chen and Lianrui Zuo and Yihao Liu",

booktitle = "Data Augmentation, Labelling, and Imperfections - 3rd MICCAI Workshop, DALI 2023 Held in Conjunction with MICCAI 2023, Proceedings",

address = "United States",

}

Lin, F, Xia, Y, Ravikumar, N, Liu, Q, MacRaild, M & Frangi, AF 2024, Adaptive Semi-supervised Segmentation of Brain Vessels with Ambiguous Labels. in Y Xue, C Chen, C Chen, L Zuo & Y Liu (eds), Data Augmentation, Labelling, and Imperfections - 3rd MICCAI Workshop, DALI 2023 Held in Conjunction with MICCAI 2023, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 14379 LNCS, Springer Nature, pp. 106-116, 3rd International Workshop on Data Augmentation, Labeling, and Imperfections, DALI 2023 in conjunction with the 26th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2023, Vancouver, Canada, 12/10/23. https://doi.org/10.48550/arXiv.2308.03613, https://doi.org/10.1007/978-3-031-58171-7_11

Adaptive Semi-supervised Segmentation of Brain Vessels with Ambiguous Labels. / Lin, Fengming; Xia, Yan; Ravikumar, Nishant et al.
Data Augmentation, Labelling, and Imperfections - 3rd MICCAI Workshop, DALI 2023 Held in Conjunction with MICCAI 2023, Proceedings. ed. / Yuan Xue; Chen Chen; Chao Chen; Lianrui Zuo; Yihao Liu. Springer Nature, 2024. p. 106-116 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14379 LNCS).

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

TY - GEN

T1 - Adaptive Semi-supervised Segmentation of Brain Vessels with Ambiguous Labels

AU - Lin, Fengming

AU - Xia, Yan

AU - Ravikumar, Nishant

AU - Liu, Qiongyao

AU - MacRaild, Michael

AU - Frangi, Alejandro F.

N1 - Publisher Copyright: © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.

PY - 2024/4/27

Y1 - 2024/4/27

N2 - Accurate segmentation of brain vessels is crucial for cerebrovascular disease diagnosis and treatment. However, existing methods face challenges in capturing small vessels and handling datasets that are partially or ambiguously annotated. In this paper, we propose an adaptive semi-supervised approach to address these challenges. Our approach incorporates innovative techniques including progressive semi-supervised learning, adaptative training strategy, and boundary enhancement. Experimental results on 3DRA datasets demonstrate the superiority of our method in terms of mesh-based segmentation metrics. By leveraging the partially and ambiguously labeled data, which only annotates the main vessels, our method achieves impressive segmentation performance on mislabeled fine vessels, showcasing its potential for clinical applications.

AB - Accurate segmentation of brain vessels is crucial for cerebrovascular disease diagnosis and treatment. However, existing methods face challenges in capturing small vessels and handling datasets that are partially or ambiguously annotated. In this paper, we propose an adaptive semi-supervised approach to address these challenges. Our approach incorporates innovative techniques including progressive semi-supervised learning, adaptative training strategy, and boundary enhancement. Experimental results on 3DRA datasets demonstrate the superiority of our method in terms of mesh-based segmentation metrics. By leveraging the partially and ambiguously labeled data, which only annotates the main vessels, our method achieves impressive segmentation performance on mislabeled fine vessels, showcasing its potential for clinical applications.

KW - Adaptive model

KW - Brain vessel segmentation

KW - Semi-supervised learning

UR - http://www.scopus.com/inward/record.url?scp=85192898917&partnerID=8YFLogxK

UR - https://www.mendeley.com/catalogue/4fbe74a9-abf8-39ce-9cae-1cf4145e3e55/

U2 - 10.48550/arXiv.2308.03613

DO - 10.48550/arXiv.2308.03613

M3 - Conference contribution

AN - SCOPUS:85192898917

SN - 9783031581700

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

SP - 106

EP - 116

BT - Data Augmentation, Labelling, and Imperfections - 3rd MICCAI Workshop, DALI 2023 Held in Conjunction with MICCAI 2023, Proceedings

A2 - Xue, Yuan

A2 - Chen, Chen

A2 - Chen, Chao

A2 - Zuo, Lianrui

A2 - Liu, Yihao

PB - Springer Nature

T2 - 3rd International Workshop on Data Augmentation, Labeling, and Imperfections, DALI 2023 in conjunction with the 26th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2023

Y2 - 12 October 2023 through 12 October 2023

ER -

Lin F, Xia Y, Ravikumar N, Liu Q, MacRaild M , Frangi AF. Adaptive Semi-supervised Segmentation of Brain Vessels with Ambiguous Labels. In Xue Y, Chen C, Chen C, Zuo L, Liu Y, editors, Data Augmentation, Labelling, and Imperfections - 3rd MICCAI Workshop, DALI 2023 Held in Conjunction with MICCAI 2023, Proceedings. Springer Nature. 2024. p. 106-116. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.48550/arXiv.2308.03613, 10.1007/978-3-031-58171-7_11