Adaptation of a Multi-Site Network to a New Clinical Site Via Batch-Normalization Similarity

Shira Kasten Serlin; Jacob Goldberger; Hayit Greenspan

doi:10.1109/isbi52829.2022.9761487

Adaptation of a Multi-Site Network to a New Clinical Site Via Batch-Normalization Similarity

Shira Kasten Serlin
, Jacob Goldberger
, Hayit Greenspan

Bar-Ilan University - The Alexander Kofkin Faculty of Engineering

Tel Aviv University

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

3 Scopus citations

Abstract

This paper tackles the challenging problem of medical site adaptation; i.e., learning a model from multi-site source data such that it can be modified and adapted to a new site using only unlabeled data from the new site. The method is based on Domain Specific Batch Normalization architecture and uses the Batch Normalization statistics of the new site to find the most similar internal site. The similarity measure is computed in an embedded space of the BN parameters. We evaluated our method on the task of MRI prostate segmentation. Public datasets from six different institutions were used, containing distribution shifts. The experimental results show that the proposed approach outperforms other generalization and adaptation methods.

Original language	English
Title of host publication	IEEE ISBI 2022 Proceedings - 2022 IEEE International Symposium on Biomedical Imaging
Publisher	IEEE Computer Society
ISBN (Electronic)	9781665429238
DOIs	https://doi.org/10.1109/isbi52829.2022.9761487
State	Published - 2022
Event	19th IEEE International Symposium on Biomedical Imaging, ISBI 2022 - Hybrid, Kolkata, India Duration: 28 Mar 2022 → 31 Mar 2022

Publication series

Name	Proceedings - International Symposium on Biomedical Imaging
Volume	2022-March
ISSN (Print)	1945-7928
ISSN (Electronic)	1945-8452

Conference

Conference	19th IEEE International Symposium on Biomedical Imaging, ISBI 2022
Country/Territory	India
City	Hybrid, Kolkata
Period	28/03/22 → 31/03/22

Bibliographical note

Publisher Copyright:
© 2022 IEEE.

Funding

This research was supported by the Ministry of Science & Technology, Israel.

Funders
Ministry of science and technology, Israel

Keywords

batch-normalization
domain adaptation
multi-site
prostate segmentation

Access to Document

10.1109/isbi52829.2022.9761487

Cite this

Kasten Serlin, S., Goldberger, J., & Greenspan, H. (2022). Adaptation of a Multi-Site Network to a New Clinical Site Via Batch-Normalization Similarity. In IEEE ISBI 2022 Proceedings - 2022 IEEE International Symposium on Biomedical Imaging (Proceedings - International Symposium on Biomedical Imaging; Vol. 2022-March). IEEE Computer Society. https://doi.org/10.1109/isbi52829.2022.9761487

@inproceedings{e1063e632c3e4529baaffc498b804287,

title = "Adaptation of a Multi-Site Network to a New Clinical Site Via Batch-Normalization Similarity",

abstract = "This paper tackles the challenging problem of medical site adaptation; i.e., learning a model from multi-site source data such that it can be modified and adapted to a new site using only unlabeled data from the new site. The method is based on Domain Specific Batch Normalization architecture and uses the Batch Normalization statistics of the new site to find the most similar internal site. The similarity measure is computed in an embedded space of the BN parameters. We evaluated our method on the task of MRI prostate segmentation. Public datasets from six different institutions were used, containing distribution shifts. The experimental results show that the proposed approach outperforms other generalization and adaptation methods.",

keywords = "batch-normalization, domain adaptation, multi-site, prostate segmentation",

author = "\{Kasten Serlin\}, Shira and Jacob Goldberger and Hayit Greenspan",

note = "Publisher Copyright: {\textcopyright} 2022 IEEE.; 19th IEEE International Symposium on Biomedical Imaging, ISBI 2022 ; Conference date: 28-03-2022 Through 31-03-2022",

year = "2022",

doi = "10.1109/isbi52829.2022.9761487",

language = "אנגלית",

series = "Proceedings - International Symposium on Biomedical Imaging",

publisher = "IEEE Computer Society",

booktitle = "IEEE ISBI 2022 Proceedings - 2022 IEEE International Symposium on Biomedical Imaging",

address = "ארצות הברית",

}

Kasten Serlin, S, Goldberger, J & Greenspan, H 2022, Adaptation of a Multi-Site Network to a New Clinical Site Via Batch-Normalization Similarity. in IEEE ISBI 2022 Proceedings - 2022 IEEE International Symposium on Biomedical Imaging. Proceedings - International Symposium on Biomedical Imaging, vol. 2022-March, IEEE Computer Society, 19th IEEE International Symposium on Biomedical Imaging, ISBI 2022, Hybrid, Kolkata, India, 28/03/22. https://doi.org/10.1109/isbi52829.2022.9761487

Adaptation of a Multi-Site Network to a New Clinical Site Via Batch-Normalization Similarity. / Kasten Serlin, Shira; Goldberger, Jacob; Greenspan, Hayit.
IEEE ISBI 2022 Proceedings - 2022 IEEE International Symposium on Biomedical Imaging. IEEE Computer Society, 2022. (Proceedings - International Symposium on Biomedical Imaging; Vol. 2022-March).

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

TY - GEN

T1 - Adaptation of a Multi-Site Network to a New Clinical Site Via Batch-Normalization Similarity

AU - Kasten Serlin, Shira

AU - Goldberger, Jacob

AU - Greenspan, Hayit

PY - 2022

Y1 - 2022

N2 - This paper tackles the challenging problem of medical site adaptation; i.e., learning a model from multi-site source data such that it can be modified and adapted to a new site using only unlabeled data from the new site. The method is based on Domain Specific Batch Normalization architecture and uses the Batch Normalization statistics of the new site to find the most similar internal site. The similarity measure is computed in an embedded space of the BN parameters. We evaluated our method on the task of MRI prostate segmentation. Public datasets from six different institutions were used, containing distribution shifts. The experimental results show that the proposed approach outperforms other generalization and adaptation methods.

AB - This paper tackles the challenging problem of medical site adaptation; i.e., learning a model from multi-site source data such that it can be modified and adapted to a new site using only unlabeled data from the new site. The method is based on Domain Specific Batch Normalization architecture and uses the Batch Normalization statistics of the new site to find the most similar internal site. The similarity measure is computed in an embedded space of the BN parameters. We evaluated our method on the task of MRI prostate segmentation. Public datasets from six different institutions were used, containing distribution shifts. The experimental results show that the proposed approach outperforms other generalization and adaptation methods.

KW - batch-normalization

KW - domain adaptation

KW - multi-site

KW - prostate segmentation

UR - https://www.scopus.com/pages/publications/85129611188

U2 - 10.1109/isbi52829.2022.9761487

DO - 10.1109/isbi52829.2022.9761487

M3 - ???researchoutput.researchoutputtypes.contributiontobookanthology.conference???

AN - SCOPUS:85129611188

T3 - Proceedings - International Symposium on Biomedical Imaging

BT - IEEE ISBI 2022 Proceedings - 2022 IEEE International Symposium on Biomedical Imaging

PB - IEEE Computer Society

T2 - 19th IEEE International Symposium on Biomedical Imaging, ISBI 2022

Y2 - 28 March 2022 through 31 March 2022

ER -

Adaptation of a Multi-Site Network to a New Clinical Site Via Batch-Normalization Similarity

Abstract

Publication series

Conference

Bibliographical note

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this