TY - GEN
T1 - Unpaired brain mr-to-ct synthesis using a structure-constrained cyclegan
AU - Yang, Heran
AU - Sun, Jian
AU - Carass, Aaron
AU - Zhao, Can
AU - Lee, Junghoon
AU - Xu, Zongben
AU - Prince, Jerry
N1 - Funding Information:
Acknowledgments. This work is supported by the NSFC (11622106, 11690011, 61721002) and the China Scholarship Council.
Publisher Copyright:
© Springer Nature Switzerland AG 2018.
PY - 2018
Y1 - 2018
N2 - The cycleGAN is becoming an influential method in medical image synthesis. However, due to a lack of direct constraints between input and synthetic images, the cycleGAN cannot guarantee structural consistency between these two images, and such consistency is of extreme importance in medical imaging. To overcome this, we propose a structure-constrained cycleGAN for brain MR-to-CT synthesis using unpaired data that defines an extra structure-consistency loss based on the modality independent neighborhood descriptor to constrain structural consistency. Additionally, we use a position-based selection strategy for selecting training images instead of a completely random selection scheme. Experimental results on synthesizing CT images from brain MR images demonstrate that our method is better than the conventional cycleGAN and approximates the cycleGAN trained with paired data.
AB - The cycleGAN is becoming an influential method in medical image synthesis. However, due to a lack of direct constraints between input and synthetic images, the cycleGAN cannot guarantee structural consistency between these two images, and such consistency is of extreme importance in medical imaging. To overcome this, we propose a structure-constrained cycleGAN for brain MR-to-CT synthesis using unpaired data that defines an extra structure-consistency loss based on the modality independent neighborhood descriptor to constrain structural consistency. Additionally, we use a position-based selection strategy for selecting training images instead of a completely random selection scheme. Experimental results on synthesizing CT images from brain MR images demonstrate that our method is better than the conventional cycleGAN and approximates the cycleGAN trained with paired data.
KW - CycleGAN
KW - Deep learning
KW - MIND
KW - MR-to-CT synthesis
UR - http://www.scopus.com/inward/record.url?scp=85057230081&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85057230081&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-00889-5_20
DO - 10.1007/978-3-030-00889-5_20
M3 - Conference contribution
AN - SCOPUS:85057230081
SN - 9783030008888
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 174
EP - 182
BT - Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support - 4th International Workshop, DLMIA 2018 and 8th International Workshop, ML-CDS 2018 Held in Conjunction with MICCAI 2018
A2 - Maier-Hein, Lena
A2 - Syeda-Mahmood, Tanveer
A2 - Taylor, Zeike
A2 - Lu, Zhi
A2 - Stoyanov, Danail
A2 - Madabhushi, Anant
A2 - Tavares, João Manuel R.S.
A2 - Nascimento, Jacinto C.
A2 - Moradi, Mehdi
A2 - Martel, Anne
A2 - Papa, Joao Paulo
A2 - Conjeti, Sailesh
A2 - Belagiannis, Vasileios
A2 - Greenspan, Hayit
A2 - Carneiro, Gustavo
A2 - Bradley, Andrew
PB - Springer Verlag
T2 - 4th International Workshop on Deep Learning in Medical Image Analysis, DLMIA 2018 and 8th International Workshop on Multimodal Learning for Clinical Decision Support, ML-CDS 2018 Held in Conjunction with MICCAI 2018
Y2 - 20 September 2018 through 20 September 2018
ER -