Self-Supervised Learning for Organs At Risk and Tumor Segmentation with Uncertainty Quantification

Ilkin Isler; Debesh Jha; Curtis Lisle; Justin Rineer; Patrick Kelly; Bulent Aydogan; Mohamed Abazeed; Damla Turgut; Ulas Bagci

doi:10.1109/ICECCME57830.2023.10252269

Self-Supervised Learning for Organs At Risk and Tumor Segmentation with Uncertainty Quantification

Ilkin Isler^*, Debesh Jha, Curtis Lisle, Justin Rineer, Patrick Kelly, Bulent Aydogan, Mohamed Abazeed, Damla Turgut, Ulas Bagci

^*Corresponding author for this work

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

1 Scopus citations

Abstract

In this study, our goal is to show the impact of self-supervised pre-training of transformers for organ at risk (OAR) and tumor segmentation as compared to costly fully-supervised learning. The proposed algorithm is called Monte Carlo Transformer based U-Net (MC-Swin-U). Unlike many other available models, our approach presents uncertainty quantification with Monte Carlo dropout strategy while generating its voxel-wise prediction. We test and validate the proposed model on both public and one private datasets and evaluate the gross tumor volume (GTV) as well as nearby risky organs' boundaries. We show that self-supervised pre-training approach improves the segmentation scores significantly while providing additional benefits for avoiding large-scale annotation costs.

Original language	English (US)
Title of host publication	International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2023
Publisher	Institute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)	9798350322972
DOIs	https://doi.org/10.1109/ICECCME57830.2023.10252269
State	Published - 2023
Event	2023 International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2023 - Tenerife, Canary Islands, Spain Duration: Jul 19 2023 → Jul 21 2023

Publication series

Name	International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2023

Conference

Conference	2023 International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2023
Country/Territory	Spain
City	Tenerife, Canary Islands
Period	7/19/23 → 7/21/23

Funding

This project is supported by the NIH funding: R01-CA246704 and R01-CA240639, and FDOH (Florida Department of Health) through the James and Esther King Biomedical Research Program-20K04.

Keywords

Deep learning
Organs at risk
Self-supervised learning
Swin Transformer
Uncertainty quantification

ASJC Scopus subject areas

Artificial Intelligence
Computer Networks and Communications
Computer Science Applications
Computer Vision and Pattern Recognition
Electrical and Electronic Engineering
Mechanical Engineering
Electronic, Optical and Magnetic Materials
Instrumentation

Access to Document

10.1109/ICECCME57830.2023.10252269

Cite this

Isler, I., Jha, D., Lisle, C., Rineer, J., Kelly, P., Aydogan, B., Abazeed, M., Turgut, D., & Bagci, U. (2023). Self-Supervised Learning for Organs At Risk and Tumor Segmentation with Uncertainty Quantification. In International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2023 (International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2023). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICECCME57830.2023.10252269

Isler, Ilkin ; Jha, Debesh ; Lisle, Curtis et al. / Self-Supervised Learning for Organs At Risk and Tumor Segmentation with Uncertainty Quantification. International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2023. Institute of Electrical and Electronics Engineers Inc., 2023. (International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2023).

@inproceedings{f70509a59dca45dcb3977b563d90df05,

title = "Self-Supervised Learning for Organs At Risk and Tumor Segmentation with Uncertainty Quantification",

abstract = "In this study, our goal is to show the impact of self-supervised pre-training of transformers for organ at risk (OAR) and tumor segmentation as compared to costly fully-supervised learning. The proposed algorithm is called Monte Carlo Transformer based U-Net (MC-Swin-U). Unlike many other available models, our approach presents uncertainty quantification with Monte Carlo dropout strategy while generating its voxel-wise prediction. We test and validate the proposed model on both public and one private datasets and evaluate the gross tumor volume (GTV) as well as nearby risky organs' boundaries. We show that self-supervised pre-training approach improves the segmentation scores significantly while providing additional benefits for avoiding large-scale annotation costs.",

keywords = "Deep learning, Organs at risk, Self-supervised learning, Swin Transformer, Uncertainty quantification",

author = "Ilkin Isler and Debesh Jha and Curtis Lisle and Justin Rineer and Patrick Kelly and Bulent Aydogan and Mohamed Abazeed and Damla Turgut and Ulas Bagci",

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; 2023 International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2023 ; Conference date: 19-07-2023 Through 21-07-2023",

year = "2023",

doi = "10.1109/ICECCME57830.2023.10252269",

language = "English (US)",

series = "International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2023",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

booktitle = "International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2023",

address = "United States",

}

Isler, I, Jha, D, Lisle, C, Rineer, J, Kelly, P, Aydogan, B, Abazeed, M, Turgut, D & Bagci, U 2023, Self-Supervised Learning for Organs At Risk and Tumor Segmentation with Uncertainty Quantification. in International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2023. International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2023, Institute of Electrical and Electronics Engineers Inc., 2023 International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2023, Tenerife, Canary Islands, Spain, 7/19/23. https://doi.org/10.1109/ICECCME57830.2023.10252269

Self-Supervised Learning for Organs At Risk and Tumor Segmentation with Uncertainty Quantification. / Isler, Ilkin; Jha, Debesh; Lisle, Curtis et al.
International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2023. Institute of Electrical and Electronics Engineers Inc., 2023. (International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2023).

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

TY - GEN

T1 - Self-Supervised Learning for Organs At Risk and Tumor Segmentation with Uncertainty Quantification

AU - Isler, Ilkin

AU - Jha, Debesh

AU - Lisle, Curtis

AU - Rineer, Justin

AU - Kelly, Patrick

AU - Aydogan, Bulent

AU - Abazeed, Mohamed

AU - Turgut, Damla

AU - Bagci, Ulas

PY - 2023

Y1 - 2023

N2 - In this study, our goal is to show the impact of self-supervised pre-training of transformers for organ at risk (OAR) and tumor segmentation as compared to costly fully-supervised learning. The proposed algorithm is called Monte Carlo Transformer based U-Net (MC-Swin-U). Unlike many other available models, our approach presents uncertainty quantification with Monte Carlo dropout strategy while generating its voxel-wise prediction. We test and validate the proposed model on both public and one private datasets and evaluate the gross tumor volume (GTV) as well as nearby risky organs' boundaries. We show that self-supervised pre-training approach improves the segmentation scores significantly while providing additional benefits for avoiding large-scale annotation costs.

AB - In this study, our goal is to show the impact of self-supervised pre-training of transformers for organ at risk (OAR) and tumor segmentation as compared to costly fully-supervised learning. The proposed algorithm is called Monte Carlo Transformer based U-Net (MC-Swin-U). Unlike many other available models, our approach presents uncertainty quantification with Monte Carlo dropout strategy while generating its voxel-wise prediction. We test and validate the proposed model on both public and one private datasets and evaluate the gross tumor volume (GTV) as well as nearby risky organs' boundaries. We show that self-supervised pre-training approach improves the segmentation scores significantly while providing additional benefits for avoiding large-scale annotation costs.

KW - Deep learning

KW - Organs at risk

KW - Self-supervised learning

KW - Swin Transformer

KW - Uncertainty quantification

UR - http://www.scopus.com/inward/record.url?scp=85174022810&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85174022810&partnerID=8YFLogxK

U2 - 10.1109/ICECCME57830.2023.10252269

DO - 10.1109/ICECCME57830.2023.10252269

M3 - Conference contribution

AN - SCOPUS:85174022810

T3 - International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2023

BT - International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2023

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2023 International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2023

Y2 - 19 July 2023 through 21 July 2023

ER -

Isler I, Jha D, Lisle C, Rineer J, Kelly P, Aydogan B et al. Self-Supervised Learning for Organs At Risk and Tumor Segmentation with Uncertainty Quantification. In International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2023. Institute of Electrical and Electronics Engineers Inc. 2023. (International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2023). doi: 10.1109/ICECCME57830.2023.10252269

Self-Supervised Learning for Organs At Risk and Tumor Segmentation with Uncertainty Quantification

Abstract

Publication series

Conference

Funding

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this