T-Net: Learning Feature Representation with Task-specific Supervision for Biomedical Image Analysis

The encoder-decoder network is widely used to learn deep feature representations from pixel-wise annotations in biomedical image analysis. Under this structure, the performance profoundly relies on the effectiveness of feature extraction achieved by the encoding network. However, few models have con...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Song, Weinan, Liang, Yuan, Yang, Jiawei, Wang, Kun, He, Lei
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Song, Weinan
Liang, Yuan
Yang, Jiawei
Wang, Kun
He, Lei
description The encoder-decoder network is widely used to learn deep feature representations from pixel-wise annotations in biomedical image analysis. Under this structure, the performance profoundly relies on the effectiveness of feature extraction achieved by the encoding network. However, few models have considered adapting the attention of the feature extractor even in different kinds of tasks. In this paper, we propose a novel training strategy by adapting the attention of the feature extractor according to different tasks for effective representation learning. Specifically, the framework, named T-Net, consists of an encoding network supervised by task-specific attention maps and a posterior network that takes in the learned features to predict the corresponding results. The attention map is obtained by the transformation from pixel-wise annotations according to the specific task, which is used as the supervision to regularize the feature extractor to focus on different locations of the recognition object. To show the effectiveness of our method, we evaluate T-Net on two different tasks, i.e. , segmentation and localization. Extensive results on three public datasets (BraTS-17, MoNuSeg and IDRiD) have indicated the effectiveness and efficiency of our proposed supervision method, especially over the conventional encoding-decoding network.
doi_str_mv 10.48550/arxiv.2002.08406
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2002_08406</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2002_08406</sourcerecordid><originalsourceid>FETCH-LOGICAL-a676-cb874ae7974565bd87414e41cbd4113c17fd691496bb0c779cd5e0a1aef6a2553</originalsourceid><addsrcrecordid>eNotz81Kw0AUBeDZuJDqA7jqvEDiTDI_ibtarBaCgs0-3Exu6sX8MZNW-_ba6upwOHDgY-xOilhlWot78N90jBMhklhkSphr5sroFecHXiD4gYY93yDMB4_8HSePAYcZZhoH_kXzBy8hfEZhQkctOb47TOiPFM5zO3r-SGOPDTno-LaHPfLVAN0pULhhVy10AW__c8HKzVO5fomKt-ftelVEYKyJXJ1ZBWhzq7TRdfPbpEIlXd0oKVMnbduYXKrc1LVw1uau0ShAArYGEq3TBVv-3V6U1eSpB3-qztrqok1_AOTiUJI</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>T-Net: Learning Feature Representation with Task-specific Supervision for Biomedical Image Analysis</title><source>arXiv.org</source><creator>Song, Weinan ; Liang, Yuan ; Yang, Jiawei ; Wang, Kun ; He, Lei</creator><creatorcontrib>Song, Weinan ; Liang, Yuan ; Yang, Jiawei ; Wang, Kun ; He, Lei</creatorcontrib><description>The encoder-decoder network is widely used to learn deep feature representations from pixel-wise annotations in biomedical image analysis. Under this structure, the performance profoundly relies on the effectiveness of feature extraction achieved by the encoding network. However, few models have considered adapting the attention of the feature extractor even in different kinds of tasks. In this paper, we propose a novel training strategy by adapting the attention of the feature extractor according to different tasks for effective representation learning. Specifically, the framework, named T-Net, consists of an encoding network supervised by task-specific attention maps and a posterior network that takes in the learned features to predict the corresponding results. The attention map is obtained by the transformation from pixel-wise annotations according to the specific task, which is used as the supervision to regularize the feature extractor to focus on different locations of the recognition object. To show the effectiveness of our method, we evaluate T-Net on two different tasks, i.e. , segmentation and localization. Extensive results on three public datasets (BraTS-17, MoNuSeg and IDRiD) have indicated the effectiveness and efficiency of our proposed supervision method, especially over the conventional encoding-decoding network.</description><identifier>DOI: 10.48550/arxiv.2002.08406</identifier><language>eng</language><subject>Computer Science - Computer Vision and Pattern Recognition</subject><creationdate>2020-02</creationdate><rights>http://creativecommons.org/licenses/by/4.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,780,885</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2002.08406$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2002.08406$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Song, Weinan</creatorcontrib><creatorcontrib>Liang, Yuan</creatorcontrib><creatorcontrib>Yang, Jiawei</creatorcontrib><creatorcontrib>Wang, Kun</creatorcontrib><creatorcontrib>He, Lei</creatorcontrib><title>T-Net: Learning Feature Representation with Task-specific Supervision for Biomedical Image Analysis</title><description>The encoder-decoder network is widely used to learn deep feature representations from pixel-wise annotations in biomedical image analysis. Under this structure, the performance profoundly relies on the effectiveness of feature extraction achieved by the encoding network. However, few models have considered adapting the attention of the feature extractor even in different kinds of tasks. In this paper, we propose a novel training strategy by adapting the attention of the feature extractor according to different tasks for effective representation learning. Specifically, the framework, named T-Net, consists of an encoding network supervised by task-specific attention maps and a posterior network that takes in the learned features to predict the corresponding results. The attention map is obtained by the transformation from pixel-wise annotations according to the specific task, which is used as the supervision to regularize the feature extractor to focus on different locations of the recognition object. To show the effectiveness of our method, we evaluate T-Net on two different tasks, i.e. , segmentation and localization. Extensive results on three public datasets (BraTS-17, MoNuSeg and IDRiD) have indicated the effectiveness and efficiency of our proposed supervision method, especially over the conventional encoding-decoding network.</description><subject>Computer Science - Computer Vision and Pattern Recognition</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2020</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotz81Kw0AUBeDZuJDqA7jqvEDiTDI_ibtarBaCgs0-3Exu6sX8MZNW-_ba6upwOHDgY-xOilhlWot78N90jBMhklhkSphr5sroFecHXiD4gYY93yDMB4_8HSePAYcZZhoH_kXzBy8hfEZhQkctOb47TOiPFM5zO3r-SGOPDTno-LaHPfLVAN0pULhhVy10AW__c8HKzVO5fomKt-ftelVEYKyJXJ1ZBWhzq7TRdfPbpEIlXd0oKVMnbduYXKrc1LVw1uau0ShAArYGEq3TBVv-3V6U1eSpB3-qztrqok1_AOTiUJI</recordid><startdate>20200219</startdate><enddate>20200219</enddate><creator>Song, Weinan</creator><creator>Liang, Yuan</creator><creator>Yang, Jiawei</creator><creator>Wang, Kun</creator><creator>He, Lei</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20200219</creationdate><title>T-Net: Learning Feature Representation with Task-specific Supervision for Biomedical Image Analysis</title><author>Song, Weinan ; Liang, Yuan ; Yang, Jiawei ; Wang, Kun ; He, Lei</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a676-cb874ae7974565bd87414e41cbd4113c17fd691496bb0c779cd5e0a1aef6a2553</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2020</creationdate><topic>Computer Science - Computer Vision and Pattern Recognition</topic><toplevel>online_resources</toplevel><creatorcontrib>Song, Weinan</creatorcontrib><creatorcontrib>Liang, Yuan</creatorcontrib><creatorcontrib>Yang, Jiawei</creatorcontrib><creatorcontrib>Wang, Kun</creatorcontrib><creatorcontrib>He, Lei</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Song, Weinan</au><au>Liang, Yuan</au><au>Yang, Jiawei</au><au>Wang, Kun</au><au>He, Lei</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>T-Net: Learning Feature Representation with Task-specific Supervision for Biomedical Image Analysis</atitle><date>2020-02-19</date><risdate>2020</risdate><abstract>The encoder-decoder network is widely used to learn deep feature representations from pixel-wise annotations in biomedical image analysis. Under this structure, the performance profoundly relies on the effectiveness of feature extraction achieved by the encoding network. However, few models have considered adapting the attention of the feature extractor even in different kinds of tasks. In this paper, we propose a novel training strategy by adapting the attention of the feature extractor according to different tasks for effective representation learning. Specifically, the framework, named T-Net, consists of an encoding network supervised by task-specific attention maps and a posterior network that takes in the learned features to predict the corresponding results. The attention map is obtained by the transformation from pixel-wise annotations according to the specific task, which is used as the supervision to regularize the feature extractor to focus on different locations of the recognition object. To show the effectiveness of our method, we evaluate T-Net on two different tasks, i.e. , segmentation and localization. Extensive results on three public datasets (BraTS-17, MoNuSeg and IDRiD) have indicated the effectiveness and efficiency of our proposed supervision method, especially over the conventional encoding-decoding network.</abstract><doi>10.48550/arxiv.2002.08406</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2002.08406
ispartof
issn
language eng
recordid cdi_arxiv_primary_2002_08406
source arXiv.org
subjects Computer Science - Computer Vision and Pattern Recognition
title T-Net: Learning Feature Representation with Task-specific Supervision for Biomedical Image Analysis
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-10T00%3A49%3A22IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=T-Net:%20Learning%20Feature%20Representation%20with%20Task-specific%20Supervision%20for%20Biomedical%20Image%20Analysis&rft.au=Song,%20Weinan&rft.date=2020-02-19&rft_id=info:doi/10.48550/arxiv.2002.08406&rft_dat=%3Carxiv_GOX%3E2002_08406%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true