TY - JOUR
T1 - TSP-UDANet
T2 - two-stage progressive unsupervised domain adaptation network for automated cross-modality cardiac segmentation
AU - Wang, Yonghui
AU - Zhang, Yifan
AU - Xu, Lisheng
AU - Qi, Shouliang
AU - Yao, Yudong
AU - Qian, Wei
AU - Greenwald, Stephen E.
AU - Qi, Lin
N1 - Publisher Copyright:
© 2023, The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature.
PY - 2023/10
Y1 - 2023/10
N2 - Accurate segmentation of cardiac anatomy is a prerequisite for the diagnosis of cardiovascular disease. However, due to differences in imaging modalities and imaging devices, known as domain shift, the segmentation performance of deep learning models lacks reliability. In this paper, we propose a two-stage progressive unsupervised domain adaptation network (TSP-UDANet) to reduce domain shift when segmenting cardiac images from various sources. We alleviate the domain shift between the feature distribution of the source and target domains by introducing an intermediate domain as a bridge. The TSP-UDANet consists of three sub-networks: a style transfer sub-network, a segmentation sub-network, and a self-training sub-network. We conduct cooperative alignment of different domains at image level, feature level, and output level. Specifically, we transform the appearance of images across domains and enhance domain invariance by adversarial learning in multiple aspects to achieve unsupervised segmentation of the target modality. We validate the TSP-UDANet on the MMWHS (unpaired MRI and CT images), MS-CMRSeg (cross-modality MRI images), and M&Ms (cross-vendor MRI images) datasets. The experimental results demonstrate excellent segmentation performance and generalizability for unlabeled target modality images.
AB - Accurate segmentation of cardiac anatomy is a prerequisite for the diagnosis of cardiovascular disease. However, due to differences in imaging modalities and imaging devices, known as domain shift, the segmentation performance of deep learning models lacks reliability. In this paper, we propose a two-stage progressive unsupervised domain adaptation network (TSP-UDANet) to reduce domain shift when segmenting cardiac images from various sources. We alleviate the domain shift between the feature distribution of the source and target domains by introducing an intermediate domain as a bridge. The TSP-UDANet consists of three sub-networks: a style transfer sub-network, a segmentation sub-network, and a self-training sub-network. We conduct cooperative alignment of different domains at image level, feature level, and output level. Specifically, we transform the appearance of images across domains and enhance domain invariance by adversarial learning in multiple aspects to achieve unsupervised segmentation of the target modality. We validate the TSP-UDANet on the MMWHS (unpaired MRI and CT images), MS-CMRSeg (cross-modality MRI images), and M&Ms (cross-vendor MRI images) datasets. The experimental results demonstrate excellent segmentation performance and generalizability for unlabeled target modality images.
KW - Cardiac segmentation
KW - Cross-modality learning
KW - Intermediate domain
KW - Unsupervised domain adaptation
UR - http://www.scopus.com/inward/record.url?scp=85170363397&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85170363397&partnerID=8YFLogxK
U2 - 10.1007/s00521-023-08939-6
DO - 10.1007/s00521-023-08939-6
M3 - Article
AN - SCOPUS:85170363397
SN - 0941-0643
VL - 35
SP - 22189
EP - 22207
JO - Neural Computing and Applications
JF - Neural Computing and Applications
IS - 30
ER -