Exploring 3D U-Net Training Configurations and Post-processing Strategies for the MICCAI 2023 Kidney and Tumor Segmentation Challenge

Kwang Hyun Uhm; Hyunjun Cho; Zhixin Xu; Seohoon Lim; Seung Won Jung; Sung Hoo Hong; Sung Jea Ko

doi:10.1007/978-3-031-54806-2_2

Exploring 3D U-Net Training Configurations and Post-processing Strategies for the MICCAI 2023 Kidney and Tumor Segmentation Challenge

Kwang Hyun Uhm
, Hyunjun Cho
, Zhixin Xu
, Seohoon Lim
, Seung Won Jung
, Sung Hoo Hong
, Sung Jea Ko^*

^*Corresponding author for this work

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

5 Citations (Scopus)

Abstract

In 2023, it is estimated that 81,800 kidney cancer cases will be newly diagnosed, and 14,890 people will die from this cancer in the United States. Preoperative dynamic contrast-enhanced abdominal computed tomography (CT) is often used for detecting lesions. However, there exists inter-observer variability due to subtle differences in the imaging features of kidney and kidney tumors. In this paper, we explore various 3D U-Net training configurations and effective post-processing strategies for accurate segmentation of kidneys, cysts, and kidney tumors in CT images. We validated our model on the dataset of the 2023 Kidney and Kidney Tumor Segmentation (KiTS23) challenge. Our method took the second place in the final ranking of KiTS23 challenge on unseen test data with an average Dice score of 0.820 and an average Surface Dice of 0.712.

Original language	English
Title of host publication	Kidney and Kidney Tumor Segmentation - MICCAI 2023 Challenge, KiTS 2023, Held in Conjunction with MICCAI 2023, Proceedings
Editors	Nicholas Heller, Andrew Wood, Christopher Weight, Fabian Isensee, Tim Rädsch, Resha Teipaul, Nikolaos Papanikolopoulos
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	8-13
Number of pages	6
ISBN (Print)	9783031548055
DOIs	https://doi.org/10.1007/978-3-031-54806-2_2
Publication status	Published - 2024
Event	3rd International Challenge on Kidney and Kidney Tumor Segmentation, KiTS 2023, which was held in conjunction with 26th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2023 - Vancouver, Canada Duration: 2023 Oct 8 → 2023 Oct 8

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	14540 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	3rd International Challenge on Kidney and Kidney Tumor Segmentation, KiTS 2023, which was held in conjunction with 26th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2023
Country/Territory	Canada
City	Vancouver
Period	23/10/8 → 23/10/8

Bibliographical note

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

UN SDGs

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

SDG 3 Good Health and Well-being

Keywords

3D U-Net
Kidney cancer
Medical image segmentation

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-031-54806-2_2

Cite this

Uhm, K. H., Cho, H., Xu, Z., Lim, S., Jung, S. W., Hong, S. H., & Ko, S. J. (2024). Exploring 3D U-Net Training Configurations and Post-processing Strategies for the MICCAI 2023 Kidney and Tumor Segmentation Challenge. In N. Heller, A. Wood, C. Weight, F. Isensee, T. Rädsch, R. Teipaul, & N. Papanikolopoulos (Eds.), Kidney and Kidney Tumor Segmentation - MICCAI 2023 Challenge, KiTS 2023, Held in Conjunction with MICCAI 2023, Proceedings (pp. 8-13). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14540 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-54806-2_2

Exploring 3D U-Net Training Configurations and Post-processing Strategies for the MICCAI 2023 Kidney and Tumor Segmentation Challenge. / Uhm, Kwang Hyun; Cho, Hyunjun; Xu, Zhixin et al.
Kidney and Kidney Tumor Segmentation - MICCAI 2023 Challenge, KiTS 2023, Held in Conjunction with MICCAI 2023, Proceedings. ed. / Nicholas Heller; Andrew Wood; Christopher Weight; Fabian Isensee; Tim Rädsch; Resha Teipaul; Nikolaos Papanikolopoulos. Springer Science and Business Media Deutschland GmbH, 2024. p. 8-13 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14540 LNCS).

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

Uhm, KH, Cho, H, Xu, Z, Lim, S, Jung, SW, Hong, SH & Ko, SJ 2024, Exploring 3D U-Net Training Configurations and Post-processing Strategies for the MICCAI 2023 Kidney and Tumor Segmentation Challenge. in N Heller, A Wood, C Weight, F Isensee, T Rädsch, R Teipaul & N Papanikolopoulos (eds), Kidney and Kidney Tumor Segmentation - MICCAI 2023 Challenge, KiTS 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. 14540 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 8-13, 3rd International Challenge on Kidney and Kidney Tumor Segmentation, KiTS 2023, which was held in conjunction with 26th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2023, Vancouver, Canada, 23/10/8. https://doi.org/10.1007/978-3-031-54806-2_2

Uhm KH, Cho H, Xu Z, Lim S, Jung SW, Hong SH et al. Exploring 3D U-Net Training Configurations and Post-processing Strategies for the MICCAI 2023 Kidney and Tumor Segmentation Challenge. In Heller N, Wood A, Weight C, Isensee F, Rädsch T, Teipaul R, Papanikolopoulos N, editors, Kidney and Kidney Tumor Segmentation - MICCAI 2023 Challenge, KiTS 2023, Held in Conjunction with MICCAI 2023, Proceedings. Springer Science and Business Media Deutschland GmbH. 2024. p. 8-13. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-54806-2_2

Uhm, Kwang Hyun ; Cho, Hyunjun ; Xu, Zhixin et al. / Exploring 3D U-Net Training Configurations and Post-processing Strategies for the MICCAI 2023 Kidney and Tumor Segmentation Challenge. Kidney and Kidney Tumor Segmentation - MICCAI 2023 Challenge, KiTS 2023, Held in Conjunction with MICCAI 2023, Proceedings. editor / Nicholas Heller ; Andrew Wood ; Christopher Weight ; Fabian Isensee ; Tim Rädsch ; Resha Teipaul ; Nikolaos Papanikolopoulos. Springer Science and Business Media Deutschland GmbH, 2024. pp. 8-13 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{c9af97bd97944291a8545ac32b42ac54,

title = "Exploring 3D U-Net Training Configurations and Post-processing Strategies for the MICCAI 2023 Kidney and Tumor Segmentation Challenge",

abstract = "In 2023, it is estimated that 81,800 kidney cancer cases will be newly diagnosed, and 14,890 people will die from this cancer in the United States. Preoperative dynamic contrast-enhanced abdominal computed tomography (CT) is often used for detecting lesions. However, there exists inter-observer variability due to subtle differences in the imaging features of kidney and kidney tumors. In this paper, we explore various 3D U-Net training configurations and effective post-processing strategies for accurate segmentation of kidneys, cysts, and kidney tumors in CT images. We validated our model on the dataset of the 2023 Kidney and Kidney Tumor Segmentation (KiTS23) challenge. Our method took the second place in the final ranking of KiTS23 challenge on unseen test data with an average Dice score of 0.820 and an average Surface Dice of 0.712.",

keywords = "3D U-Net, Kidney cancer, Medical image segmentation",

author = "Uhm, \{Kwang Hyun\} and Hyunjun Cho and Zhixin Xu and Seohoon Lim and Jung, \{Seung Won\} and Hong, \{Sung Hoo\} and Ko, \{Sung Jea\}",

note = "Publisher Copyright: {\textcopyright} The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.; 3rd International Challenge on Kidney and Kidney Tumor Segmentation, KiTS 2023, which was held in conjunction with 26th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2023 ; Conference date: 08-10-2023 Through 08-10-2023",

year = "2024",

doi = "10.1007/978-3-031-54806-2\_2",

language = "English",

isbn = "9783031548055",

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 = "8--13",

editor = "Nicholas Heller and Andrew Wood and Christopher Weight and Fabian Isensee and Tim R{\"a}dsch and Resha Teipaul and Nikolaos Papanikolopoulos",

booktitle = "Kidney and Kidney Tumor Segmentation - MICCAI 2023 Challenge, KiTS 2023, Held in Conjunction with MICCAI 2023, Proceedings",

address = "Germany",

}

TY - GEN

T1 - Exploring 3D U-Net Training Configurations and Post-processing Strategies for the MICCAI 2023 Kidney and Tumor Segmentation Challenge

AU - Uhm, Kwang Hyun

AU - Cho, Hyunjun

AU - Xu, Zhixin

AU - Lim, Seohoon

AU - Jung, Seung Won

AU - Hong, Sung Hoo

AU - Ko, Sung Jea

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

PY - 2024

Y1 - 2024

N2 - In 2023, it is estimated that 81,800 kidney cancer cases will be newly diagnosed, and 14,890 people will die from this cancer in the United States. Preoperative dynamic contrast-enhanced abdominal computed tomography (CT) is often used for detecting lesions. However, there exists inter-observer variability due to subtle differences in the imaging features of kidney and kidney tumors. In this paper, we explore various 3D U-Net training configurations and effective post-processing strategies for accurate segmentation of kidneys, cysts, and kidney tumors in CT images. We validated our model on the dataset of the 2023 Kidney and Kidney Tumor Segmentation (KiTS23) challenge. Our method took the second place in the final ranking of KiTS23 challenge on unseen test data with an average Dice score of 0.820 and an average Surface Dice of 0.712.

AB - In 2023, it is estimated that 81,800 kidney cancer cases will be newly diagnosed, and 14,890 people will die from this cancer in the United States. Preoperative dynamic contrast-enhanced abdominal computed tomography (CT) is often used for detecting lesions. However, there exists inter-observer variability due to subtle differences in the imaging features of kidney and kidney tumors. In this paper, we explore various 3D U-Net training configurations and effective post-processing strategies for accurate segmentation of kidneys, cysts, and kidney tumors in CT images. We validated our model on the dataset of the 2023 Kidney and Kidney Tumor Segmentation (KiTS23) challenge. Our method took the second place in the final ranking of KiTS23 challenge on unseen test data with an average Dice score of 0.820 and an average Surface Dice of 0.712.

KW - 3D U-Net

KW - Kidney cancer

KW - Medical image segmentation

UR - https://www.scopus.com/pages/publications/85188733317

U2 - 10.1007/978-3-031-54806-2_2

DO - 10.1007/978-3-031-54806-2_2

M3 - Conference contribution

AN - SCOPUS:85188733317

SN - 9783031548055

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

SP - 8

EP - 13

BT - Kidney and Kidney Tumor Segmentation - MICCAI 2023 Challenge, KiTS 2023, Held in Conjunction with MICCAI 2023, Proceedings

A2 - Heller, Nicholas

A2 - Wood, Andrew

A2 - Weight, Christopher

A2 - Isensee, Fabian

A2 - Rädsch, Tim

A2 - Teipaul, Resha

A2 - Papanikolopoulos, Nikolaos

PB - Springer Science and Business Media Deutschland GmbH

T2 - 3rd International Challenge on Kidney and Kidney Tumor Segmentation, KiTS 2023, which was held in conjunction with 26th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2023

Y2 - 8 October 2023 through 8 October 2023

ER -