Cross-modality image synthesis from unpaired data using CycleGAN: Effects of gradient consistency loss and training data size

By Yuta Hiasa, Y. Otake, M. Takao, Takumi Matsuoka, K. Takashima, Jerry L Prince, N. Sugano, Yoshinobu Sato. 2018

232 citations recorded by the source corpus.

CT is commonly used in orthopedic procedures. MRI is used along with CT to identify muscle structures and diagnose osteonecrosis due to its superior soft tissue contrast. However, MRI has poor contrast for bone structures. Clearly, it would be helpful if a corresponding CT were available, as bone boundaries are more clearly seen and CT has standardized (i.e., Hounsfield) units. Therefore, we aim at MR-to-CT synthesis. The CycleGAN was successfully applied to unpaired CT and MR images of the head, these images do not have as much variation of intensity pairs as do images in the pelvic region due to the presence of joints and muscles. In this paper, we extended the CycleGAN approach by adding the gradient consistency loss to improve the accuracy at the boundaries. We conducted two experiments. To evaluate image synthesis, we investigated dependency of image synthesis accuracy on 1) the number of training data and 2) the gradient consistency loss. To demonstrate the applicability of our method, we also investigated a segmentation accuracy on synthesized images.

Subfields: Computer Graphics and Computer-Aided Design, Computer Vision and Pattern Recognition, Generative Models, Medical Imaging

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Graphics and Computer-Aided Design

1. Aykut Erdem
2. Erkut Erdem

In Computer Vision and Pattern Recognition

1. Aykut Erdem
2. Shekhar S. Chandra
3. Stuart Crozier
4. Erkut Erdem
5. Paul S. Babyn
6. P. Greer
7. Xin Yi
8. Salman Ul Hassan Dar

In Generative Models

1. Aykut Erdem
2. Erkut Erdem
3. Paul S. Babyn
4. Xin Yi
5. Levent Karacan
6. Mahmut Yurt
7. Ekta Walia
8. Tolga Cukur

In Medical Imaging

1. Paul S. Babyn
2. Xin Yi
3. Ekta Walia

Citations in the corpus

• Generative Adversarial Network in Medical Imaging: A Review (2018, background): “Hiasa et al. (2018) further incorporated gradient consistency loss in the training to improve accuracy at the boundaries. Zhang et al. (2018c) found that using only cycle loss in the cross modality”
• Manipulating Medical Image Translation with Manifold Disentanglement (2020, methodology): “When pair-wise labels are not available, CycleGAN [29] and UNIT [30] are used for semi-supervised learning on unpaired images [31], [32], [33], [34], [35].”
• Image Synthesis in Multi-Contrast MRI With Conditional Generative Adversarial Networks (2018, methodology): “Recently, cross-modality synthesis with GANs was leveraged as a pre-processing step to enhance various medical imaging tasks such as segmentation, classification or tumor detection [29], [33], [36], [39], [40], [79], [80].”