An UNet-Based Brain Tumor Segmentation Framework via Optimal Mass Transportation Pre-processing

Jia Wei Liao; Tsung Ming Huang; Tiexiang Li; Wen Wei Lin; Han Wang; Shing Tung Yau

doi:10.1007/978-3-031-33842-7_19

An UNet-Based Brain Tumor Segmentation Framework via Optimal Mass Transportation Pre-processing

Jia Wei Liao, Tsung Ming Huang^*, Tiexiang Li, Wen Wei Lin, Han Wang, Shing Tung Yau

^*Corresponding author for this work

Department of Mathematics

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

1 Citation (Scopus)

Abstract

This article aims to build a framework for MRI images of brain tumor segmentation using the deep learning method. For this purpose, we develop a novel 2-Phase UNet-based OMT framework to increase the ratio of brain tumors using optimal mass transportation (OMT). Moreover, due to the scarcity of training data, we change the density function by different parameters to increase the data diversity. For the post-processing, we propose an adaptive ensemble procedure by solving the eigenvectors of the Dice similarity matrix and choosing the result with the highest aggregation probability as the predicted label. The Dice scores of the whole tumor (WT), tumor core (TC), and enhanced tumor (ET) regions for online validation computed by SegResUNet were 0.9214, 0.8823, and 0.8411, respectively. Compared with random crop pre-processing, OMT is far superior.

Original language	English
Title of host publication	Brainlesion
Subtitle of host publication	Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 8th International Workshop, BrainLes 2022, Held in Conjunction with MICCAI 2022, Revised Selected Papers
Editors	Spyridon Bakas, Ujjwal Baid, Bhakti Baheti, Alessandro Crimi, Sylwia Malec, Monika Pytlarz, Maximilian Zenk, Reuben Dorent
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	216-228
Number of pages	13
ISBN (Print)	9783031338410
DOIs	https://doi.org/10.1007/978-3-031-33842-7_19
Publication status	Published - 2023
Event	Proceedings of the 8th International MICCAI Brainlesion Workshop, BrainLes 2022 - Singapore, Singapore Duration: 2022 Sept 18 → 2022 Sept 22

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	13769 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	Proceedings of the 8th International MICCAI Brainlesion Workshop, BrainLes 2022
Country/Territory	Singapore
City	Singapore
Period	2022/09/18 → 2022/09/22

Keywords

2-Phase OMT framework
Brain Tumor Segmentation
Density function
Optimal Mass Transportation
UNet

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-031-33842-7_19

Cite this

Liao, J. W., Huang, T. M., Li, T., Lin, W. W., Wang, H., & Yau, S. T. (2023). An UNet-Based Brain Tumor Segmentation Framework via Optimal Mass Transportation Pre-processing. In S. Bakas, U. Baid, B. Baheti, A. Crimi, S. Malec, M. Pytlarz, M. Zenk, & R. Dorent (Eds.), Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 8th International Workshop, BrainLes 2022, Held in Conjunction with MICCAI 2022, Revised Selected Papers (pp. 216-228). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13769 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-33842-7_19

An UNet-Based Brain Tumor Segmentation Framework via Optimal Mass Transportation Pre-processing. / Liao, Jia Wei; Huang, Tsung Ming; Li, Tiexiang et al.
Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 8th International Workshop, BrainLes 2022, Held in Conjunction with MICCAI 2022, Revised Selected Papers. ed. / Spyridon Bakas; Ujjwal Baid; Bhakti Baheti; Alessandro Crimi; Sylwia Malec; Monika Pytlarz; Maximilian Zenk; Reuben Dorent. Springer Science and Business Media Deutschland GmbH, 2023. p. 216-228 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13769 LNCS).

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

Liao, JW, Huang, TM, Li, T, Lin, WW, Wang, H & Yau, ST 2023, An UNet-Based Brain Tumor Segmentation Framework via Optimal Mass Transportation Pre-processing. in S Bakas, U Baid, B Baheti, A Crimi, S Malec, M Pytlarz, M Zenk & R Dorent (eds), Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 8th International Workshop, BrainLes 2022, Held in Conjunction with MICCAI 2022, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 13769 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 216-228, Proceedings of the 8th International MICCAI Brainlesion Workshop, BrainLes 2022, Singapore, Singapore, 2022/09/18. https://doi.org/10.1007/978-3-031-33842-7_19

Liao JW, Huang TM, Li T, Lin WW, Wang H, Yau ST. An UNet-Based Brain Tumor Segmentation Framework via Optimal Mass Transportation Pre-processing. In Bakas S, Baid U, Baheti B, Crimi A, Malec S, Pytlarz M, Zenk M, Dorent R, editors, Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 8th International Workshop, BrainLes 2022, Held in Conjunction with MICCAI 2022, Revised Selected Papers. Springer Science and Business Media Deutschland GmbH. 2023. p. 216-228. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-33842-7_19

Liao, Jia Wei ; Huang, Tsung Ming ; Li, Tiexiang et al. / An UNet-Based Brain Tumor Segmentation Framework via Optimal Mass Transportation Pre-processing. Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 8th International Workshop, BrainLes 2022, Held in Conjunction with MICCAI 2022, Revised Selected Papers. editor / Spyridon Bakas ; Ujjwal Baid ; Bhakti Baheti ; Alessandro Crimi ; Sylwia Malec ; Monika Pytlarz ; Maximilian Zenk ; Reuben Dorent. Springer Science and Business Media Deutschland GmbH, 2023. pp. 216-228 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{a46f46549d7a45b7b8e63316d8c53643,

title = "An UNet-Based Brain Tumor Segmentation Framework via Optimal Mass Transportation Pre-processing",

abstract = "This article aims to build a framework for MRI images of brain tumor segmentation using the deep learning method. For this purpose, we develop a novel 2-Phase UNet-based OMT framework to increase the ratio of brain tumors using optimal mass transportation (OMT). Moreover, due to the scarcity of training data, we change the density function by different parameters to increase the data diversity. For the post-processing, we propose an adaptive ensemble procedure by solving the eigenvectors of the Dice similarity matrix and choosing the result with the highest aggregation probability as the predicted label. The Dice scores of the whole tumor (WT), tumor core (TC), and enhanced tumor (ET) regions for online validation computed by SegResUNet were 0.9214, 0.8823, and 0.8411, respectively. Compared with random crop pre-processing, OMT is far superior.",

keywords = "2-Phase OMT framework, Brain Tumor Segmentation, Density function, Optimal Mass Transportation, UNet",

author = "Liao, {Jia Wei} and Huang, {Tsung Ming} and Tiexiang Li and Lin, {Wen Wei} and Han Wang and Yau, {Shing Tung}",

note = "Publisher Copyright: {\textcopyright} 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.; Proceedings of the 8th International MICCAI Brainlesion Workshop, BrainLes 2022 ; Conference date: 18-09-2022 Through 22-09-2022",

year = "2023",

doi = "10.1007/978-3-031-33842-7_19",

language = "English",

isbn = "9783031338410",

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 = "216--228",

editor = "Spyridon Bakas and Ujjwal Baid and Bhakti Baheti and Alessandro Crimi and Sylwia Malec and Monika Pytlarz and Maximilian Zenk and Reuben Dorent",

booktitle = "Brainlesion",

}

TY - GEN

T1 - An UNet-Based Brain Tumor Segmentation Framework via Optimal Mass Transportation Pre-processing

AU - Liao, Jia Wei

AU - Huang, Tsung Ming

AU - Li, Tiexiang

AU - Lin, Wen Wei

AU - Wang, Han

AU - Yau, Shing Tung

PY - 2023

Y1 - 2023

N2 - This article aims to build a framework for MRI images of brain tumor segmentation using the deep learning method. For this purpose, we develop a novel 2-Phase UNet-based OMT framework to increase the ratio of brain tumors using optimal mass transportation (OMT). Moreover, due to the scarcity of training data, we change the density function by different parameters to increase the data diversity. For the post-processing, we propose an adaptive ensemble procedure by solving the eigenvectors of the Dice similarity matrix and choosing the result with the highest aggregation probability as the predicted label. The Dice scores of the whole tumor (WT), tumor core (TC), and enhanced tumor (ET) regions for online validation computed by SegResUNet were 0.9214, 0.8823, and 0.8411, respectively. Compared with random crop pre-processing, OMT is far superior.

AB - This article aims to build a framework for MRI images of brain tumor segmentation using the deep learning method. For this purpose, we develop a novel 2-Phase UNet-based OMT framework to increase the ratio of brain tumors using optimal mass transportation (OMT). Moreover, due to the scarcity of training data, we change the density function by different parameters to increase the data diversity. For the post-processing, we propose an adaptive ensemble procedure by solving the eigenvectors of the Dice similarity matrix and choosing the result with the highest aggregation probability as the predicted label. The Dice scores of the whole tumor (WT), tumor core (TC), and enhanced tumor (ET) regions for online validation computed by SegResUNet were 0.9214, 0.8823, and 0.8411, respectively. Compared with random crop pre-processing, OMT is far superior.

KW - 2-Phase OMT framework

KW - Brain Tumor Segmentation

KW - Density function

KW - Optimal Mass Transportation

KW - UNet

UR - http://www.scopus.com/inward/record.url?scp=85172287341&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85172287341&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-33842-7_19

DO - 10.1007/978-3-031-33842-7_19

M3 - Conference contribution

AN - SCOPUS:85172287341

SN - 9783031338410

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

SP - 216

EP - 228

BT - Brainlesion

A2 - Bakas, Spyridon

A2 - Baid, Ujjwal

A2 - Baheti, Bhakti

A2 - Crimi, Alessandro

A2 - Malec, Sylwia

A2 - Pytlarz, Monika

A2 - Zenk, Maximilian

A2 - Dorent, Reuben

PB - Springer Science and Business Media Deutschland GmbH

T2 - Proceedings of the 8th International MICCAI Brainlesion Workshop, BrainLes 2022

Y2 - 18 September 2022 through 22 September 2022

ER -