Title:
画像分割方法、装置及びコンピュータプログラム
Document Type and Number:
Japanese Patent JP7268248
Kind Code:
B2
Abstract:
Embodiments of the present disclosure disclose an image segmentation method and apparatus and a storage medium. In the embodiments of the present disclosure, a target domain image and a source domain image that is labeled with target information are obtained first, then the source domain image and the target domain image are segmented by using a generative network in a first generative adversarial network and a generative network in a second generative adversarial network, next, a first source domain target image and a second source domain target image are determined according to a first source domain segmentation loss and a second source domain segmentation loss, a first target domain target image and a second target domain target image are determined according to a first target domain segmentation loss and a second target domain segmentation loss, then cross training is performed on the first generative adversarial network and the second generative adversarial network to obtain the trained first generative adversarial network, and a to-be-segmented image is then segmented based on the generative network in the trained first generative adversarial network to obtain a segmentation result.
Inventors:
柳 露▲艷▼
▲馬▼ ▲カイ▼
▲鄭▼ 冶▲楓▼
▲馬▼ ▲カイ▼
▲鄭▼ 冶▲楓▼
Application Number:
JP2022523505A
Publication Date:
May 02, 2023
Filing Date:
October 29, 2020
Export Citation:
Assignee:
TENCENT TECHNOLOGY(SHENZHEN)COMPANY LIMITED
International Classes:
G06T7/00; G06N3/04; G06N3/08; G06V10/82
Domestic Patent References:
| JP2019207491A |
Foreign References:
| CN109345455A | ||||
| CN109255390A | ||||
| US20190220977 | ||||
| WO2018200072A1 | ||||
| WO2020028382A1 |
Other References:
Junlin Yang et al.,"Domain-Agnostic Learning With Anatomy-Consistent Embedding for Cross-Modality Liver Segmentation",2019 IEEE/CVF International Conference on Computer Vision Workshop (ICCVW),米国,IEEE,2019年10月27日,pp.323-331
Chenjie Ge et al.,"Cross-Modality Augmentation of Brain MR Images Using a Novel Pairwise Generative Adversarial Network for Enhanced Glioma Classification",2019 IEEE International Conference on Image Processing (ICIP),米国,IEEE,2019年08月26日,pp.559-563
Zizhao Zhang et al.,"Translating and Segmenting Multimodal Medical Volumes with Cycle- and Shape-Consistency Generative Adversarial Network",2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition,米国,IEEE,2018年06月18日,pp.9242-9251
Bo Han et al.,"Co-teaching: Robust Training of Deep Neural Networks with Extremely Noisy Labels",arXiv,米国,Cornell University,2018年10月30日,pp.1-13,https://arxiv.org/abs/1804.06872
Chenjie Ge et al.,"Cross-Modality Augmentation of Brain MR Images Using a Novel Pairwise Generative Adversarial Network for Enhanced Glioma Classification",2019 IEEE International Conference on Image Processing (ICIP),米国,IEEE,2019年08月26日,pp.559-563
Zizhao Zhang et al.,"Translating and Segmenting Multimodal Medical Volumes with Cycle- and Shape-Consistency Generative Adversarial Network",2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition,米国,IEEE,2018年06月18日,pp.9242-9251
Bo Han et al.,"Co-teaching: Robust Training of Deep Neural Networks with Extremely Noisy Labels",arXiv,米国,Cornell University,2018年10月30日,pp.1-13,https://arxiv.org/abs/1804.06872
Attorney, Agent or Firm:
Shinya Mihiro
Naoki Matsuo
Naoki Matsuo