Multi-Contrast MRI Segmentation Trained on Synthetic Images

Ismail Irmakci; Zeki Emre Unel; Nazli Ikizler-Cinbis; Ulas Bagci

doi:10.1109/EMBC48229.2022.9871119

Multi-Contrast MRI Segmentation Trained on Synthetic Images

Ismail Irmakci, Zeki Emre Unel, Nazli Ikizler-Cinbis, Ulas Bagci^*

^*Corresponding author for this work

Radiology

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

3 Scopus citations

Abstract

In our comprehensive experiments and evaluations, we show that it is possible to generate multiple contrast (even all synthetically) and use synthetically generated images to train an image segmentation engine. We showed promising segmentation results tested on real multi-contrast MRI scans when delineating muscle, fat, bone and bone marrow, all trained on synthetic images. Based on synthetic image training, our segmentation results were as high as 93.91%, 94.11%, 91.63%, 95.33%, for muscle, fat, bone, and bone marrow delineation, respectively. Results were not significantly different from the ones obtained when real images were used for segmentation training: 94.68%, 94.67%, 95.91%, and 96.82%, respectively. Clinical relevance - Synthetically generated images could potentially be used in large-scale training of deep networks for segmentation purpose. Small data set problem of many clinical imaging problems can potentially be addressed with the proposed algorithm.

Original language	English (US)
Title of host publication	44th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBC 2022
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	5030-5034
Number of pages	5
ISBN (Electronic)	9781728127828
DOIs	https://doi.org/10.1109/EMBC48229.2022.9871119
State	Published - 2022
Event	44th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBC 2022 - Glasgow, United Kingdom Duration: Jul 11 2022 → Jul 15 2022

Publication series

Name	Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS
Volume	2022-July
ISSN (Print)	1557-170X

Conference

Conference	44th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBC 2022
Country/Territory	United Kingdom
City	Glasgow
Period	7/11/22 → 7/15/22

Funding

Acknowledgments Our study is exempt from human subject section as the data is publicly available and fully anonymized. This study is approved under the existing IRB at Baltimore Longitudinal Study of Aging (BLSA) [12]. This study is partially supported by the NIH grant R01-CA246704-01 and R01-CA240639-01. We thank Ege University for letting us to use their servers for running our experiments.

ASJC Scopus subject areas

Signal Processing
Health Informatics
Computer Vision and Pattern Recognition
Biomedical Engineering

Access to Document

10.1109/EMBC48229.2022.9871119

Cite this

Irmakci, I., Unel, Z. E., Ikizler-Cinbis, N., & Bagci, U. (2022). Multi-Contrast MRI Segmentation Trained on Synthetic Images. In 44th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBC 2022 (pp. 5030-5034). (Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS; Vol. 2022-July). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/EMBC48229.2022.9871119

Irmakci, Ismail ; Unel, Zeki Emre ; Ikizler-Cinbis, Nazli et al. / Multi-Contrast MRI Segmentation Trained on Synthetic Images. 44th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBC 2022. Institute of Electrical and Electronics Engineers Inc., 2022. pp. 5030-5034 (Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS).

@inproceedings{2eb13117566846a7993ce26ea11bcdca,

title = "Multi-Contrast MRI Segmentation Trained on Synthetic Images",

abstract = "In our comprehensive experiments and evaluations, we show that it is possible to generate multiple contrast (even all synthetically) and use synthetically generated images to train an image segmentation engine. We showed promising segmentation results tested on real multi-contrast MRI scans when delineating muscle, fat, bone and bone marrow, all trained on synthetic images. Based on synthetic image training, our segmentation results were as high as 93.91%, 94.11%, 91.63%, 95.33%, for muscle, fat, bone, and bone marrow delineation, respectively. Results were not significantly different from the ones obtained when real images were used for segmentation training: 94.68%, 94.67%, 95.91%, and 96.82%, respectively. Clinical relevance - Synthetically generated images could potentially be used in large-scale training of deep networks for segmentation purpose. Small data set problem of many clinical imaging problems can potentially be addressed with the proposed algorithm.",

author = "Ismail Irmakci and Unel, {Zeki Emre} and Nazli Ikizler-Cinbis and Ulas Bagci",

note = "Publisher Copyright: {\textcopyright} 2022 IEEE.; 44th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBC 2022 ; Conference date: 11-07-2022 Through 15-07-2022",

year = "2022",

doi = "10.1109/EMBC48229.2022.9871119",

language = "English (US)",

series = "Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS",

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

pages = "5030--5034",

booktitle = "44th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBC 2022",

address = "United States",

}

Irmakci, I, Unel, ZE, Ikizler-Cinbis, N & Bagci, U 2022, Multi-Contrast MRI Segmentation Trained on Synthetic Images. in 44th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBC 2022. Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS, vol. 2022-July, Institute of Electrical and Electronics Engineers Inc., pp. 5030-5034, 44th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBC 2022, Glasgow, United Kingdom, 7/11/22. https://doi.org/10.1109/EMBC48229.2022.9871119

Multi-Contrast MRI Segmentation Trained on Synthetic Images. / Irmakci, Ismail; Unel, Zeki Emre; Ikizler-Cinbis, Nazli et al.
44th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBC 2022. Institute of Electrical and Electronics Engineers Inc., 2022. p. 5030-5034 (Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS; Vol. 2022-July).

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

TY - GEN

T1 - Multi-Contrast MRI Segmentation Trained on Synthetic Images

AU - Irmakci, Ismail

AU - Unel, Zeki Emre

AU - Ikizler-Cinbis, Nazli

AU - Bagci, Ulas

PY - 2022

Y1 - 2022

N2 - In our comprehensive experiments and evaluations, we show that it is possible to generate multiple contrast (even all synthetically) and use synthetically generated images to train an image segmentation engine. We showed promising segmentation results tested on real multi-contrast MRI scans when delineating muscle, fat, bone and bone marrow, all trained on synthetic images. Based on synthetic image training, our segmentation results were as high as 93.91%, 94.11%, 91.63%, 95.33%, for muscle, fat, bone, and bone marrow delineation, respectively. Results were not significantly different from the ones obtained when real images were used for segmentation training: 94.68%, 94.67%, 95.91%, and 96.82%, respectively. Clinical relevance - Synthetically generated images could potentially be used in large-scale training of deep networks for segmentation purpose. Small data set problem of many clinical imaging problems can potentially be addressed with the proposed algorithm.

AB - In our comprehensive experiments and evaluations, we show that it is possible to generate multiple contrast (even all synthetically) and use synthetically generated images to train an image segmentation engine. We showed promising segmentation results tested on real multi-contrast MRI scans when delineating muscle, fat, bone and bone marrow, all trained on synthetic images. Based on synthetic image training, our segmentation results were as high as 93.91%, 94.11%, 91.63%, 95.33%, for muscle, fat, bone, and bone marrow delineation, respectively. Results were not significantly different from the ones obtained when real images were used for segmentation training: 94.68%, 94.67%, 95.91%, and 96.82%, respectively. Clinical relevance - Synthetically generated images could potentially be used in large-scale training of deep networks for segmentation purpose. Small data set problem of many clinical imaging problems can potentially be addressed with the proposed algorithm.

UR - http://www.scopus.com/inward/record.url?scp=85138128560&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85138128560&partnerID=8YFLogxK

U2 - 10.1109/EMBC48229.2022.9871119

DO - 10.1109/EMBC48229.2022.9871119

M3 - Conference contribution

C2 - 36086321

AN - SCOPUS:85138128560

T3 - Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS

SP - 5030

EP - 5034

BT - 44th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBC 2022

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 44th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBC 2022

Y2 - 11 July 2022 through 15 July 2022

ER -

Irmakci I, Unel ZE, Ikizler-Cinbis N, Bagci U. Multi-Contrast MRI Segmentation Trained on Synthetic Images. In 44th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBC 2022. Institute of Electrical and Electronics Engineers Inc. 2022. p. 5030-5034. (Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS). doi: 10.1109/EMBC48229.2022.9871119

Multi-Contrast MRI Segmentation Trained on Synthetic Images

Abstract

Publication series

Conference

Funding

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this