An Ensemble of 3D U-Net Based Models for Segmentation of Kidney and Masses in CT Scans

Alex Golts; Daniel Khapun; Daniel Shats; Yoel Shoshan; Flora Gilboa-Solomon

doi:10.1007/978-3-030-98385-7_14

An Ensemble of 3D U-Net Based Models for Segmentation of Kidney and Masses in CT Scans

Alex Golts
, Daniel Khapun
, Daniel Shats
, Yoel Shoshan
, Flora Gilboa-Solomon

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

Abstract

Automatic segmentation of renal tumors and surrounding anatomy in computed tomography (CT) scans is a promising tool for assisting radiologists and surgeons in their efforts to study these scans and improve the prospect of treating kidney cancer. We describe our approach, which we used to compete in the 2021 Kidney and Kidney Tumor Segmentation (KiTS21) challenge. Our approach is based on the successful 3D U-Net architecture with our added innovations, including the use of transfer learning, an unsupervised regularized loss, custom postprocessing, and multi-annotator ground truth that mimics the evaluation protocol. Our submission has reached the 2nd place in the KiTS21 challenge.

Original language	English
Title of host publication	Kidney and Kidney Tumor Segmentation - MICCAI 2021 Challenge, KiTS 2021, Held in Conjunction with MICCAI 2021, Proceedings
Editors	Nicholas Heller, Fabian Isensee, Darya Trofimova, Resha Tejpaul, Nikolaos Papanikolopoulos, Christopher Weight
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	103-115
Number of pages	13
ISBN (Print)	9783030983840
DOIs	https://doi.org/10.1007/978-3-030-98385-7_14
State	Published - 2022
Externally published	Yes
Event	2nd International challenge on Kidney and Kidney Tumor Segmentation, KiTS 2021 held in conjunction with 24th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2021 - Virtual, Online, France Duration: 27 Sep 2021 → 27 Sep 2021

Publication series

Name	Lecture Notes in Computer Science
Volume	13168 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	2nd International challenge on Kidney and Kidney Tumor Segmentation, KiTS 2021 held in conjunction with 24th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2021
Country/Territory	France
City	Virtual, Online
Period	27/09/21 → 27/09/21

Bibliographical note

Publisher Copyright:
© 2022, Springer Nature Switzerland AG.

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 tumor
Medical imaging
Semantic segmentation

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-030-98385-7_14

Cite this

Golts, A., Khapun, D., Shats, D., Shoshan, Y., & Gilboa-Solomon, F. (2022). An Ensemble of 3D U-Net Based Models for Segmentation of Kidney and Masses in CT Scans. In N. Heller, F. Isensee, D. Trofimova, R. Tejpaul, N. Papanikolopoulos, & C. Weight (Eds.), Kidney and Kidney Tumor Segmentation - MICCAI 2021 Challenge, KiTS 2021, Held in Conjunction with MICCAI 2021, Proceedings (pp. 103-115). (Lecture Notes in Computer Science; Vol. 13168 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-030-98385-7_14

Golts, Alex ; Khapun, Daniel ; Shats, Daniel et al. / An Ensemble of 3D U-Net Based Models for Segmentation of Kidney and Masses in CT Scans. Kidney and Kidney Tumor Segmentation - MICCAI 2021 Challenge, KiTS 2021, Held in Conjunction with MICCAI 2021, Proceedings. editor / Nicholas Heller ; Fabian Isensee ; Darya Trofimova ; Resha Tejpaul ; Nikolaos Papanikolopoulos ; Christopher Weight. Springer Science and Business Media Deutschland GmbH, 2022. pp. 103-115 (Lecture Notes in Computer Science).

@inproceedings{da944f394d26437eb4998e122c260449,

title = "An Ensemble of 3D U-Net Based Models for Segmentation of Kidney and Masses in CT Scans",

abstract = "Automatic segmentation of renal tumors and surrounding anatomy in computed tomography (CT) scans is a promising tool for assisting radiologists and surgeons in their efforts to study these scans and improve the prospect of treating kidney cancer. We describe our approach, which we used to compete in the 2021 Kidney and Kidney Tumor Segmentation (KiTS21) challenge. Our approach is based on the successful 3D U-Net architecture with our added innovations, including the use of transfer learning, an unsupervised regularized loss, custom postprocessing, and multi-annotator ground truth that mimics the evaluation protocol. Our submission has reached the 2nd place in the KiTS21 challenge.",

keywords = "3D U-Net, Kidney tumor, Medical imaging, Semantic segmentation",

author = "Alex Golts and Daniel Khapun and Daniel Shats and Yoel Shoshan and Flora Gilboa-Solomon",

note = "Publisher Copyright: {\textcopyright} 2022, Springer Nature Switzerland AG.; 2nd International challenge on Kidney and Kidney Tumor Segmentation, KiTS 2021 held in conjunction with 24th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2021 ; Conference date: 27-09-2021 Through 27-09-2021",

year = "2022",

doi = "10.1007/978-3-030-98385-7\_14",

language = "English",

isbn = "9783030983840",

series = "Lecture Notes in Computer Science",

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

pages = "103--115",

editor = "Nicholas Heller and Fabian Isensee and Darya Trofimova and Resha Tejpaul and Nikolaos Papanikolopoulos and Christopher Weight",

booktitle = "Kidney and Kidney Tumor Segmentation - MICCAI 2021 Challenge, KiTS 2021, Held in Conjunction with MICCAI 2021, Proceedings",

address = "Germany",

}

Golts, A, Khapun, D, Shats, D, Shoshan, Y & Gilboa-Solomon, F 2022, An Ensemble of 3D U-Net Based Models for Segmentation of Kidney and Masses in CT Scans. in N Heller, F Isensee, D Trofimova, R Tejpaul, N Papanikolopoulos & C Weight (eds), Kidney and Kidney Tumor Segmentation - MICCAI 2021 Challenge, KiTS 2021, Held in Conjunction with MICCAI 2021, Proceedings. Lecture Notes in Computer Science, vol. 13168 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 103-115, 2nd International challenge on Kidney and Kidney Tumor Segmentation, KiTS 2021 held in conjunction with 24th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2021, Virtual, Online, France, 27/09/21. https://doi.org/10.1007/978-3-030-98385-7_14

An Ensemble of 3D U-Net Based Models for Segmentation of Kidney and Masses in CT Scans. / Golts, Alex; Khapun, Daniel; Shats, Daniel et al.
Kidney and Kidney Tumor Segmentation - MICCAI 2021 Challenge, KiTS 2021, Held in Conjunction with MICCAI 2021, Proceedings. ed. / Nicholas Heller; Fabian Isensee; Darya Trofimova; Resha Tejpaul; Nikolaos Papanikolopoulos; Christopher Weight. Springer Science and Business Media Deutschland GmbH, 2022. p. 103-115 (Lecture Notes in Computer Science; Vol. 13168 LNCS).

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

TY - GEN

T1 - An Ensemble of 3D U-Net Based Models for Segmentation of Kidney and Masses in CT Scans

AU - Golts, Alex

AU - Khapun, Daniel

AU - Shats, Daniel

AU - Shoshan, Yoel

AU - Gilboa-Solomon, Flora

PY - 2022

Y1 - 2022

N2 - Automatic segmentation of renal tumors and surrounding anatomy in computed tomography (CT) scans is a promising tool for assisting radiologists and surgeons in their efforts to study these scans and improve the prospect of treating kidney cancer. We describe our approach, which we used to compete in the 2021 Kidney and Kidney Tumor Segmentation (KiTS21) challenge. Our approach is based on the successful 3D U-Net architecture with our added innovations, including the use of transfer learning, an unsupervised regularized loss, custom postprocessing, and multi-annotator ground truth that mimics the evaluation protocol. Our submission has reached the 2nd place in the KiTS21 challenge.

AB - Automatic segmentation of renal tumors and surrounding anatomy in computed tomography (CT) scans is a promising tool for assisting radiologists and surgeons in their efforts to study these scans and improve the prospect of treating kidney cancer. We describe our approach, which we used to compete in the 2021 Kidney and Kidney Tumor Segmentation (KiTS21) challenge. Our approach is based on the successful 3D U-Net architecture with our added innovations, including the use of transfer learning, an unsupervised regularized loss, custom postprocessing, and multi-annotator ground truth that mimics the evaluation protocol. Our submission has reached the 2nd place in the KiTS21 challenge.

KW - 3D U-Net

KW - Kidney tumor

KW - Medical imaging

KW - Semantic segmentation

UR - https://www.scopus.com/pages/publications/85127674220

U2 - 10.1007/978-3-030-98385-7_14

DO - 10.1007/978-3-030-98385-7_14

M3 - Conference contribution

AN - SCOPUS:85127674220

SN - 9783030983840

T3 - Lecture Notes in Computer Science

SP - 103

EP - 115

BT - Kidney and Kidney Tumor Segmentation - MICCAI 2021 Challenge, KiTS 2021, Held in Conjunction with MICCAI 2021, Proceedings

A2 - Heller, Nicholas

A2 - Isensee, Fabian

A2 - Trofimova, Darya

A2 - Tejpaul, Resha

A2 - Papanikolopoulos, Nikolaos

A2 - Weight, Christopher

PB - Springer Science and Business Media Deutschland GmbH

T2 - 2nd International challenge on Kidney and Kidney Tumor Segmentation, KiTS 2021 held in conjunction with 24th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2021

Y2 - 27 September 2021 through 27 September 2021

ER -

Golts A, Khapun D, Shats D, Shoshan Y, Gilboa-Solomon F. An Ensemble of 3D U-Net Based Models for Segmentation of Kidney and Masses in CT Scans. In Heller N, Isensee F, Trofimova D, Tejpaul R, Papanikolopoulos N, Weight C, editors, Kidney and Kidney Tumor Segmentation - MICCAI 2021 Challenge, KiTS 2021, Held in Conjunction with MICCAI 2021, Proceedings. Springer Science and Business Media Deutschland GmbH. 2022. p. 103-115. (Lecture Notes in Computer Science). doi: 10.1007/978-3-030-98385-7_14