Unsupervised Domain Adaptation Network With Category-Centric Prototype Aligner for Biomedical Image Segmentation

With the widespread success of deep learning in biomedical image segmentation, domain shift becomes a critical and challenging problem, as the gap between two domains can severely affect model performance when deployed to unseen data with heterogeneous features. To alleviate this problem, we present...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Ping Gong, Wenwen Yu, Qiuwen Sun, Ruohan Zhao, Junfeng Hu
Formato:	article
Lenguaje:	EN
Publicado:	IEEE 2021
Materias:	Biomedical image segmentation cross-modality learning unsupervised domain adaptation category-centric prototype aligner Electrical engineering. Electronics. Nuclear engineering TK1-9971
Acceso en línea:	https://doaj.org/article/e246d9dba5aa49b89dd40c7392e666f7
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:e246d9dba5aa49b89dd40c7392e666f7
record_format	dspace
spelling	oai:doaj.org-article:e246d9dba5aa49b89dd40c7392e666f72021-11-19T00:06:46ZUnsupervised Domain Adaptation Network With Category-Centric Prototype Aligner for Biomedical Image Segmentation2169-353610.1109/ACCESS.2021.3063634https://doaj.org/article/e246d9dba5aa49b89dd40c7392e666f72021-01-01T00:00:00Zhttps://ieeexplore.ieee.org/document/9367129/https://doaj.org/toc/2169-3536With the widespread success of deep learning in biomedical image segmentation, domain shift becomes a critical and challenging problem, as the gap between two domains can severely affect model performance when deployed to unseen data with heterogeneous features. To alleviate this problem, we present a novel unsupervised domain adaptation network, for generalizing models learned from the labeled source domain to the unlabeled target domain for cross-modality biomedical image segmentation. Specifically, our approach consists of two key modules, a conditional domain discriminator (CDD) and a category-centric prototype aligner (CCPA). The CDD, extended from conditional domain adversarial networks in classifier tasks, is effective and robust in handling complex cross-modality biomedical images. The CCPA, improved from the graph-induced prototype alignment mechanism in cross-domain object detection, can exploit precise instance-level features through an elaborate prototype representation. In addition, it can address the negative effect of class imbalance via entropy-based loss. Extensive experiments on a public benchmark for the cardiac substructure segmentation task demonstrate that our method significantly improves performance on the target domain.Ping GongWenwen YuQiuwen SunRuohan ZhaoJunfeng HuIEEEarticleBiomedical image segmentationcross-modality learningunsupervised domain adaptationcategory-centric prototype alignerElectrical engineering. Electronics. Nuclear engineeringTK1-9971ENIEEE Access, Vol 9, Pp 36500-36511 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Biomedical image segmentation cross-modality learning unsupervised domain adaptation category-centric prototype aligner Electrical engineering. Electronics. Nuclear engineering TK1-9971
spellingShingle	Biomedical image segmentation cross-modality learning unsupervised domain adaptation category-centric prototype aligner Electrical engineering. Electronics. Nuclear engineering TK1-9971 Ping Gong Wenwen Yu Qiuwen Sun Ruohan Zhao Junfeng Hu Unsupervised Domain Adaptation Network With Category-Centric Prototype Aligner for Biomedical Image Segmentation
description	With the widespread success of deep learning in biomedical image segmentation, domain shift becomes a critical and challenging problem, as the gap between two domains can severely affect model performance when deployed to unseen data with heterogeneous features. To alleviate this problem, we present a novel unsupervised domain adaptation network, for generalizing models learned from the labeled source domain to the unlabeled target domain for cross-modality biomedical image segmentation. Specifically, our approach consists of two key modules, a conditional domain discriminator (CDD) and a category-centric prototype aligner (CCPA). The CDD, extended from conditional domain adversarial networks in classifier tasks, is effective and robust in handling complex cross-modality biomedical images. The CCPA, improved from the graph-induced prototype alignment mechanism in cross-domain object detection, can exploit precise instance-level features through an elaborate prototype representation. In addition, it can address the negative effect of class imbalance via entropy-based loss. Extensive experiments on a public benchmark for the cardiac substructure segmentation task demonstrate that our method significantly improves performance on the target domain.
format	article
author	Ping Gong Wenwen Yu Qiuwen Sun Ruohan Zhao Junfeng Hu
author_facet	Ping Gong Wenwen Yu Qiuwen Sun Ruohan Zhao Junfeng Hu
author_sort	Ping Gong
title	Unsupervised Domain Adaptation Network With Category-Centric Prototype Aligner for Biomedical Image Segmentation
title_short	Unsupervised Domain Adaptation Network With Category-Centric Prototype Aligner for Biomedical Image Segmentation
title_full	Unsupervised Domain Adaptation Network With Category-Centric Prototype Aligner for Biomedical Image Segmentation
title_fullStr	Unsupervised Domain Adaptation Network With Category-Centric Prototype Aligner for Biomedical Image Segmentation
title_full_unstemmed	Unsupervised Domain Adaptation Network With Category-Centric Prototype Aligner for Biomedical Image Segmentation
title_sort	unsupervised domain adaptation network with category-centric prototype aligner for biomedical image segmentation
publisher	IEEE
publishDate	2021
url	https://doaj.org/article/e246d9dba5aa49b89dd40c7392e666f7
work_keys_str_mv	AT pinggong unsuperviseddomainadaptationnetworkwithcategorycentricprototypealignerforbiomedicalimagesegmentation AT wenwenyu unsuperviseddomainadaptationnetworkwithcategorycentricprototypealignerforbiomedicalimagesegmentation AT qiuwensun unsuperviseddomainadaptationnetworkwithcategorycentricprototypealignerforbiomedicalimagesegmentation AT ruohanzhao unsuperviseddomainadaptationnetworkwithcategorycentricprototypealignerforbiomedicalimagesegmentation AT junfenghu unsuperviseddomainadaptationnetworkwithcategorycentricprototypealignerforbiomedicalimagesegmentation
_version_	1718420641257881600

Unsupervised Domain Adaptation Network With Category-Centric Prototype Aligner for Biomedical Image Segmentation

Ejemplares similares