An Asymmetric Cycle-Consistency Loss For Dealing With Many-To-One Mappings In Image Translation: A Study On Thigh Mr Scans

M. Gadermayr; M. Tschuchnig; L. Gupta; N. Krämer; D. Truhn; D. Merhof; B. Gess

doi:10.1109/ISBI48211.2021.9433891

An Asymmetric Cycle-Consistency Loss For Dealing With Many-To-One Mappings In Image Translation: A Study On Thigh Mr Scans

M. Gadermayr
, M. Tschuchnig
, L. Gupta
, N. Krämer
, D. Truhn
, D. Merhof
, B. Gess

Institute of Imaging & Computer Vision, RWTH Aachen University
University of Salzburg
Department of Radiology, University Hospital Aachen, RWTH Aachen University
Department of Neurology, University Hospital Aachen, RWTH Aachen University

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

Abstract

Adversarial networks using a cycle-consistency loss facilitate unpaired training of image-translation models and thereby exhibit a high potential in medical applications. However, the fact that images in one domain potentially map to more than one image in another domain (e.g. in case of pathological changes) exhibits a major challenge for training the networks. We offer a solution to improve the training process in case of many-to-one mappings by modifying the cycle-consistency loss. We show formally and empirically that the proposed method improves the performance without radically changing the architecture and increasing the model complexity. © 2021 IEEE.

Original language	English
Title of host publication	2021 IEEE 18th International Symposium on Biomedical Imaging (ISBI)
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	1182-1186
Number of pages	5
ISBN (Electronic)	978-1-6654-1246-9
ISBN (Print)	978-1-6654-2947-4
DOIs	https://doi.org/10.1109/ISBI48211.2021.9433891
Publication status	Published - 2021
Event	18th IEEE International Symposium on Biomedical Imaging, ISBI 2021 - Duration: 13 Apr 2021 → 16 Apr 2021 https://biomedicalimaging.org/2021/

Conference

Conference	18th IEEE International Symposium on Biomedical Imaging, ISBI 2021
Abbreviated title	ISBI 2021
Period	13/04/21 → 16/04/21
Internet address	https://biomedicalimaging.org/2021/

UN SDGs

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

SDG 3 Good Health and Well-being
SDG 9 Industry, Innovation, and Infrastructure

Keywords

Mapping
Medical applications
Adversarial networks
High potential
Image translation
Many-to-one
Model complexity
Pathological changes
Training process
Medical imaging

Classification according to Österreichische Systematik der Wissenschaftszweige (ÖFOS 2012)

102003 Image processing

Applied Research Level (ARL)

ARL Level 4 - Experimental setup in laboratory-like conditions

Research focus/foci

Industrial Informatics
Applied Health Innovation

Access to Document

10.1109/ISBI48211.2021.9433891

https://www.scopus.com/inward/record.uri?eid=2-s2.0-85107222747&doi=10.1109%2fISBI48211.2021.9433891&partnerID=40&md5=a73e3b755c571a6ce83533f72f7ede04

KIAMed: Artificial Intelligence for the Analysis of Medical Imaging Data
Gadermayr, M. (PI), Oostingh, G. J. (CoPI) & Tschuchnig, M. E. (CoI)
1/01/20 → 31/10/23
Project: Funded research

Cite this

Gadermayr, M., Tschuchnig, M., Gupta, L., Krämer, N., Truhn, D., Merhof, D., & Gess, B. (2021). An Asymmetric Cycle-Consistency Loss For Dealing With Many-To-One Mappings In Image Translation: A Study On Thigh Mr Scans. In 2021 IEEE 18th International Symposium on Biomedical Imaging (ISBI) (pp. 1182-1186). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ISBI48211.2021.9433891

@inproceedings{02a25d82efff40d7b318095be6c6c7ac,

title = "An Asymmetric Cycle-Consistency Loss For Dealing With Many-To-One Mappings In Image Translation: A Study On Thigh Mr Scans",

abstract = "Adversarial networks using a cycle-consistency loss facilitate unpaired training of image-translation models and thereby exhibit a high potential in medical applications. However, the fact that images in one domain potentially map to more than one image in another domain (e.g. in case of pathological changes) exhibits a major challenge for training the networks. We offer a solution to improve the training process in case of many-to-one mappings by modifying the cycle-consistency loss. We show formally and empirically that the proposed method improves the performance without radically changing the architecture and increasing the model complexity. {\textcopyright} 2021 IEEE.",

keywords = "Mapping, Medical applications, Adversarial networks, High potential, Image translation, Many-to-one, Model complexity, Pathological changes, Training process, Medical imaging",

author = "M. Gadermayr and M. Tschuchnig and L. Gupta and N. Kr{\"a}mer and D. Truhn and D. Merhof and B. Gess",

note = "Conference code: 169147 Cited By :3 Export Date: 14 December 2023 References: Zhu, J., Park, T., Isola, P., Efros, A.A., Unpaired image-to-image translation using cycle-consistent adversarial networks (2017) Proceedings of the International Conference on Computer Vision (ICCV'17); Yi, Z., Zhang, H., Tan, P., Gong, M., Dualgan: Unsupervised dual learning for image-to-image translation (2017) Proceedings of the IEEE International Conference on Computer Vision (ICCV'17); Wolterink, J.M., Dinkla, A.M., Savenije, F.M.H., Seevinck, P.R., Van Den Berg, T.C.A., Isgum, I., Deep MR to CT synthesis using unpaired data (2017) Proceedings of the International MICCAI Workshop Simulation and Synthesis in Medical Imaging (SASHIMF17), pp. 14-23; Jiang, J., Hu, Y., Tyagi, N., Zhang, P., Rimner, A., Mageras, G.S., Deasy, J.O., Veeraraghavan, H., Tumor-aware, adversarial domain adaptation from CT to MRI for lung cancer segmentation (2018) Proceedings of the Conference on Medical Image Computing and Computer Assisted Intervention (MICCAF18), p. 777785; Bentaieb, A., Hamarneh, G., Adversarial stain transfer for histopathology image analysis (2018) IEEE Transactions on Medical Imaging, 37 (3), pp. 792-802; Zhu, J., Zhang, R., Pathak, D., Darrell, T., Efros, A.A., Wang, O., Shechtman, E., Toward multimodal image-to-image translation (2017) Advances in Neural Information Processing Systems (NIPS'17), pp. 465-476; Almahairi, A., Rajeshwar, S., Sor-Doni, A., Bachman, P., Courville, A.C., Augmented cyclegan: Learning many-to-many mappings from unpaired data (2018) Proceedings of International Conference on Machine Learning (ICML'18); Huang, X., Liu, M., Belongie, S., Kautz, J., Multimodal unsupervised image-to-image translation (2018) Proceedings of the European Conference on Computer Vision (ECCV'18); Park, T., Efros, A.A., Zhang, R., Zhu, J., Contrastive learning for conditional image synthesis (2020) Proceedings of the European Conference on Computer Vision (ECCV'20); Gao, Y., Liu, Y., Wang, Y., Shi, Z., Yu, J., A universal intensity standardization method based on a many-to-one weak-paired cycle generative adversarial network for magnetic resonance images (2019) IEEE Transactions on Medical Imaging, 38 (9), pp. 2059-2069. , Sept; Gadermayr, M., Disch, C., Miiller, M., Merhof, D., Gess, B., A comprehensive study on automated muscle segmentation for assessing fat infiltration in neuromuscular diseases (2018) Magnetic Reconance Imaging, 48, pp. 20-26; Yao, J., Kovacs, W., Hsieh, N., Liu, C., Summers, R.M., Holistic segmentation of intermuscular adipose tissues on thigh MRI (2017) Proceedings of the International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAF17), p. 737745; Ronneberger, O., Fischer, P., Brox, T., U-net: Convolutional networks for biomedical image segmentation (2015) Proceedings of the International Conference on Medical Image Computing and Computer Aided Interventions (MIC-CAF15), pp. 234-241; 18th IEEE International Symposium on Biomedical Imaging, ISBI 2021, ISBI 2021 ; Conference date: 13-04-2021 Through 16-04-2021",

year = "2021",

doi = "10.1109/ISBI48211.2021.9433891",

language = "English",

isbn = "978-1-6654-2947-4",

pages = "1182--1186",

booktitle = "2021 IEEE 18th International Symposium on Biomedical Imaging (ISBI)",

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

address = "United States",

url = "https://biomedicalimaging.org/2021/",

}

Gadermayr, M, Tschuchnig, M, Gupta, L, Krämer, N, Truhn, D, Merhof, D & Gess, B 2021, An Asymmetric Cycle-Consistency Loss For Dealing With Many-To-One Mappings In Image Translation: A Study On Thigh Mr Scans. in 2021 IEEE 18th International Symposium on Biomedical Imaging (ISBI). Institute of Electrical and Electronics Engineers Inc., pp. 1182-1186, 18th IEEE International Symposium on Biomedical Imaging, ISBI 2021, 13/04/21. https://doi.org/10.1109/ISBI48211.2021.9433891

An Asymmetric Cycle-Consistency Loss For Dealing With Many-To-One Mappings In Image Translation: A Study On Thigh Mr Scans. / Gadermayr, M.; Tschuchnig, M.; Gupta, L. et al.
2021 IEEE 18th International Symposium on Biomedical Imaging (ISBI). Institute of Electrical and Electronics Engineers Inc., 2021. p. 1182-1186.

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

TY - GEN

T1 - An Asymmetric Cycle-Consistency Loss For Dealing With Many-To-One Mappings In Image Translation: A Study On Thigh Mr Scans

AU - Gadermayr, M.

AU - Tschuchnig, M.

AU - Gupta, L.

AU - Krämer, N.

AU - Truhn, D.

AU - Merhof, D.

AU - Gess, B.

N1 - Conference code: 169147 Cited By :3 Export Date: 14 December 2023 References: Zhu, J., Park, T., Isola, P., Efros, A.A., Unpaired image-to-image translation using cycle-consistent adversarial networks (2017) Proceedings of the International Conference on Computer Vision (ICCV'17); Yi, Z., Zhang, H., Tan, P., Gong, M., Dualgan: Unsupervised dual learning for image-to-image translation (2017) Proceedings of the IEEE International Conference on Computer Vision (ICCV'17); Wolterink, J.M., Dinkla, A.M., Savenije, F.M.H., Seevinck, P.R., Van Den Berg, T.C.A., Isgum, I., Deep MR to CT synthesis using unpaired data (2017) Proceedings of the International MICCAI Workshop Simulation and Synthesis in Medical Imaging (SASHIMF17), pp. 14-23; Jiang, J., Hu, Y., Tyagi, N., Zhang, P., Rimner, A., Mageras, G.S., Deasy, J.O., Veeraraghavan, H., Tumor-aware, adversarial domain adaptation from CT to MRI for lung cancer segmentation (2018) Proceedings of the Conference on Medical Image Computing and Computer Assisted Intervention (MICCAF18), p. 777785; Bentaieb, A., Hamarneh, G., Adversarial stain transfer for histopathology image analysis (2018) IEEE Transactions on Medical Imaging, 37 (3), pp. 792-802; Zhu, J., Zhang, R., Pathak, D., Darrell, T., Efros, A.A., Wang, O., Shechtman, E., Toward multimodal image-to-image translation (2017) Advances in Neural Information Processing Systems (NIPS'17), pp. 465-476; Almahairi, A., Rajeshwar, S., Sor-Doni, A., Bachman, P., Courville, A.C., Augmented cyclegan: Learning many-to-many mappings from unpaired data (2018) Proceedings of International Conference on Machine Learning (ICML'18); Huang, X., Liu, M., Belongie, S., Kautz, J., Multimodal unsupervised image-to-image translation (2018) Proceedings of the European Conference on Computer Vision (ECCV'18); Park, T., Efros, A.A., Zhang, R., Zhu, J., Contrastive learning for conditional image synthesis (2020) Proceedings of the European Conference on Computer Vision (ECCV'20); Gao, Y., Liu, Y., Wang, Y., Shi, Z., Yu, J., A universal intensity standardization method based on a many-to-one weak-paired cycle generative adversarial network for magnetic resonance images (2019) IEEE Transactions on Medical Imaging, 38 (9), pp. 2059-2069. , Sept; Gadermayr, M., Disch, C., Miiller, M., Merhof, D., Gess, B., A comprehensive study on automated muscle segmentation for assessing fat infiltration in neuromuscular diseases (2018) Magnetic Reconance Imaging, 48, pp. 20-26; Yao, J., Kovacs, W., Hsieh, N., Liu, C., Summers, R.M., Holistic segmentation of intermuscular adipose tissues on thigh MRI (2017) Proceedings of the International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAF17), p. 737745; Ronneberger, O., Fischer, P., Brox, T., U-net: Convolutional networks for biomedical image segmentation (2015) Proceedings of the International Conference on Medical Image Computing and Computer Aided Interventions (MIC-CAF15), pp. 234-241

PY - 2021

Y1 - 2021

N2 - Adversarial networks using a cycle-consistency loss facilitate unpaired training of image-translation models and thereby exhibit a high potential in medical applications. However, the fact that images in one domain potentially map to more than one image in another domain (e.g. in case of pathological changes) exhibits a major challenge for training the networks. We offer a solution to improve the training process in case of many-to-one mappings by modifying the cycle-consistency loss. We show formally and empirically that the proposed method improves the performance without radically changing the architecture and increasing the model complexity. © 2021 IEEE.

AB - Adversarial networks using a cycle-consistency loss facilitate unpaired training of image-translation models and thereby exhibit a high potential in medical applications. However, the fact that images in one domain potentially map to more than one image in another domain (e.g. in case of pathological changes) exhibits a major challenge for training the networks. We offer a solution to improve the training process in case of many-to-one mappings by modifying the cycle-consistency loss. We show formally and empirically that the proposed method improves the performance without radically changing the architecture and increasing the model complexity. © 2021 IEEE.

KW - Mapping

KW - Medical applications

KW - Adversarial networks

KW - High potential

KW - Image translation

KW - Many-to-one

KW - Model complexity

KW - Pathological changes

KW - Training process

KW - Medical imaging

U2 - 10.1109/ISBI48211.2021.9433891

DO - 10.1109/ISBI48211.2021.9433891

M3 - Conference contribution

SN - 978-1-6654-2947-4

SP - 1182

EP - 1186

BT - 2021 IEEE 18th International Symposium on Biomedical Imaging (ISBI)

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 18th IEEE International Symposium on Biomedical Imaging, ISBI 2021

Y2 - 13 April 2021 through 16 April 2021

ER -

Gadermayr M, Tschuchnig M, Gupta L, Krämer N, Truhn D, Merhof D et al. An Asymmetric Cycle-Consistency Loss For Dealing With Many-To-One Mappings In Image Translation: A Study On Thigh Mr Scans. In 2021 IEEE 18th International Symposium on Biomedical Imaging (ISBI). Institute of Electrical and Electronics Engineers Inc. 2021. p. 1182-1186 doi: 10.1109/ISBI48211.2021.9433891

An Asymmetric Cycle-Consistency Loss For Dealing With Many-To-One Mappings In Image Translation: A Study On Thigh Mr Scans

Abstract

Conference

UN SDGs

Keywords

Classification according to Österreichische Systematik der Wissenschaftszweige (ÖFOS 2012)

Applied Research Level (ARL)

Research focus/foci

Access to Document

Fingerprint

Projects

KIAMed: Artificial Intelligence for the Analysis of Medical Imaging Data

Cite this