南方科技大学知识苑(SUSTech KC): Unseen Domain Generalization for Prostate MRI Segmentation via Disentangled Representations

题名	Unseen Domain Generalization for Prostate MRI Segmentation via Disentangled Representations
作者	Lu，Ye 1; Xing，Xiaohan 1; Meng，Max Q.H.1,2,3
DOI	10.1109/ROBIO54168.2021.9739504
发表日期	2021
ISBN	978-1-6654-0536-2
会议录名称	2021 IEEE International Conference on Robotics and Biomimetics, ROBIO 2021
页码	1986-1991
会议日期	27-31 Dec. 2021
会议地点	Sanya, China
摘要	In clinical practice, medical images obtained from different sites often exhibit appearance variations, resulting in limited generalizability of deep learning models for segmentation in deployment. It is an important but challenging task to train a model which can directly generalize to unseen domains with distribution shifts. In this paper, we propose to disentangle content from style representations for prostate MRI segmentation to improve the model generalization, considering anatomical content information is domain invariant and decides the segmentation masks. Our method roots in a representation disentanglement network, sharing the content encoder with the segmentation module to remove the effect of appearance discrepancy. Besides, we introduce two domain discriminators to further regularize the disentangled representation learning. We extensively validate our model on a multi-site dataset for prostate MRI segmentation. Both quantitative and qualitative experimental results demonstrate the effectiveness of our method, outperforming the baseline method and many state-of-the-art generalization methods.
关键词	Representation learning Deep learning Image segmentation Image analysis Magnetic resonance imaging Conferences Biomimetics
学校署名	其他
语种	英语
相关链接	[Scopus记录]
收录类别	EI
EI入藏号	20221611977597
EI主题词	Deep learning ; Image segmentation ; Medical imaging
EI分类号	Biomedical Engineering:461.1 ; Ergonomics and Human Factors Engineering:461.4 ; Medicine and Pharmacology:461.6 ; Imaging Techniques:746
Scopus记录号	2-s2.0-85128208135
来源库	Scopus
全文链接	https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9739504
引用统计	被引频次[WOS]：2
成果类型	会议论文
条目标识符	http://sustech.caswiz.com/handle/2SGJ60CL/331183
专题	工学院_电子与电气工程系
作者单位	1.Chinese University of Hong Kong,Department of Electronic Engineering,Hong Kong,Hong Kong 2.Southern University of Science and Technology,Department of Electronic and Electrical Engineering,Shenzhen,China 3.Shenzhen Research Institute,Chinese University of Hong Kong,Shenzhen,China
推荐引用方式 GB/T 7714	Lu，Ye,Xing，Xiaohan,Meng，Max Q.H.. Unseen Domain Generalization for Prostate MRI Segmentation via Disentangled Representations[C],2021:1986-1991.