No Cover Image

Journal article 59 views 4 downloads

MFR‐UNet: A Medical Image Segmentation Network With Fused Multi‐Scale Feature Refinement

Shaoqiang Wang Orcid Logo, Guiling Shi, Shuo Sun, Yuchen Wang, Yulin Zhang Orcid Logo, Weixian Li, Yawu Zhao, Cheng Cheng Orcid Logo

IET Systems Biology, Volume: 20, Issue: 1

Swansea University Author: Cheng Cheng Orcid Logo

  • 71127.VoR.pdf

    PDF | Version of Record

    © 2025 The Author(s). This is an open access article under the terms of the Creative Commons Attribution License.

    Download (1.69MB)

Check full text

DOI (Published version): 10.1049/syb2.70049

Abstract

Medical image segmentation is crucial for clinical diagnosis and treatment planning. Although methods based on CNN, particularly U-Net and its variants, have achieved remarkable success in automated segmentation tasks, they still face challenges in effectively capturing long-range dependencies, refi...

Full description

Published in: IET Systems Biology
ISSN: 1751-8849 1751-8857
Published: Institution of Engineering and Technology (IET) 2026
Online Access: Check full text

URI: https://cronfa.swan.ac.uk/Record/cronfa71127
first_indexed 2025-12-10T12:22:57Z
last_indexed 2026-01-08T05:22:01Z
id cronfa71127
recordtype SURis
fullrecord <?xml version="1.0"?><rfc1807><datestamp>2026-01-07T17:14:57.2390185</datestamp><bib-version>v2</bib-version><id>71127</id><entry>2025-12-10</entry><title>MFR&#x2010;UNet: A Medical Image Segmentation Network With Fused Multi&#x2010;Scale Feature Refinement</title><swanseaauthors><author><sid>11ddf61c123b99e59b00fa1479367582</sid><ORCID>0000-0003-0371-9646</ORCID><firstname>Cheng</firstname><surname>Cheng</surname><name>Cheng Cheng</name><active>true</active><ethesisStudent>false</ethesisStudent></author></swanseaauthors><date>2025-12-10</date><deptcode>MACS</deptcode><abstract>Medical image segmentation is crucial for clinical diagnosis and treatment planning. Although methods based on CNN, particularly U-Net and its variants, have achieved remarkable success in automated segmentation tasks, they still face challenges in effectively capturing long-range dependencies, refining multi-level features, and efficiently integrating cross-level information. To address these issues, we propose a novel U-Net architecture incorporating a multi-scale feature refinement mechanism (MFR-UNet). This network enhances segmentation accuracy and robustness by integrating three innovative modules. First, we designed a wavelet transform convolution (WtConv) module. By decomposing, processing, and reconstructing features in the frequency domain, this module enables the model to learn high-frequency details and low-frequency contours with greater precision. Second, we introduce a large receptive field attention (LRFA) module in the encoder. Combining deep separable convolutions with multi-head attention, LRFA efficiently captures global contextual information at low computational cost. Finally, in the skip connections and decoding path, our weighted contextual fusion module (WCF) module dynamically generates channel attention weights for one feature stream to another, achieving efficient adaptive feature fusion. Simulation experiments on multiple public medical image segmentation datasets demonstrate that our MFR-UNet outperforms several existing mainstream methods in key metrics such as Dice coefficient and IoU, proving its effectiveness in enhancing segmentation accuracy and boundary clarity.</abstract><type>Journal Article</type><journal>IET Systems Biology</journal><volume>20</volume><journalNumber>1</journalNumber><paginationStart/><paginationEnd/><publisher>Institution of Engineering and Technology (IET)</publisher><placeOfPublication/><isbnPrint/><isbnElectronic/><issnPrint>1751-8849</issnPrint><issnElectronic>1751-8857</issnElectronic><keywords>feature fusion; large receptive field; medical image segmentation; wavelet</keywords><publishedDay>1</publishedDay><publishedMonth>1</publishedMonth><publishedYear>2026</publishedYear><publishedDate>2026-01-01</publishedDate><doi>10.1049/syb2.70049</doi><url/><notes/><college>COLLEGE NANME</college><department>Mathematics and Computer Science School</department><CollegeCode>COLLEGE CODE</CollegeCode><DepartmentCode>MACS</DepartmentCode><institution>Swansea University</institution><apcterm>SU Library paid the OA fee (TA Institutional Deal)</apcterm><funders>Doctoral Training Centre at Swansea University (Grant Number: RS718); UKRI (Grant Number: EP/W020408/1); Hebei Province Education Science &#x201D;14th Five-Year Plan&#x201D; (Grant Number: 2303065)</funders><projectreference/><lastEdited>2026-01-07T17:14:57.2390185</lastEdited><Created>2025-12-10T12:16:49.7341739</Created><path><level id="1">Faculty of Science and Engineering</level><level id="2">School of Mathematics and Computer Science - Computer Science</level></path><authors><author><firstname>Shaoqiang</firstname><surname>Wang</surname><orcid>0000-0002-7539-5970</orcid><order>1</order></author><author><firstname>Guiling</firstname><surname>Shi</surname><order>2</order></author><author><firstname>Shuo</firstname><surname>Sun</surname><order>3</order></author><author><firstname>Yuchen</firstname><surname>Wang</surname><order>4</order></author><author><firstname>Yulin</firstname><surname>Zhang</surname><orcid>0000-0002-9125-5273</orcid><order>5</order></author><author><firstname>Weixian</firstname><surname>Li</surname><order>6</order></author><author><firstname>Yawu</firstname><surname>Zhao</surname><order>7</order></author><author><firstname>Cheng</firstname><surname>Cheng</surname><orcid>0000-0003-0371-9646</orcid><order>8</order></author></authors><documents><document><filename>71127__35915__d68c22c5d2884147b900afebd9d27ecc.pdf</filename><originalFilename>71127.VoR.pdf</originalFilename><uploaded>2026-01-07T17:12:49.2584562</uploaded><type>Output</type><contentLength>1770955</contentLength><contentType>application/pdf</contentType><version>Version of Record</version><cronfaStatus>true</cronfaStatus><documentNotes>&#xA9; 2025 The Author(s). This is an open access article under the terms of the Creative Commons Attribution License.</documentNotes><copyrightCorrect>true</copyrightCorrect><language>eng</language><licence>http://creativecommons.org/licenses/by/4.0/</licence></document></documents><OutputDurs/></rfc1807>
spelling 2026-01-07T17:14:57.2390185 v2 71127 2025-12-10 MFR‐UNet: A Medical Image Segmentation Network With Fused Multi‐Scale Feature Refinement 11ddf61c123b99e59b00fa1479367582 0000-0003-0371-9646 Cheng Cheng Cheng Cheng true false 2025-12-10 MACS Medical image segmentation is crucial for clinical diagnosis and treatment planning. Although methods based on CNN, particularly U-Net and its variants, have achieved remarkable success in automated segmentation tasks, they still face challenges in effectively capturing long-range dependencies, refining multi-level features, and efficiently integrating cross-level information. To address these issues, we propose a novel U-Net architecture incorporating a multi-scale feature refinement mechanism (MFR-UNet). This network enhances segmentation accuracy and robustness by integrating three innovative modules. First, we designed a wavelet transform convolution (WtConv) module. By decomposing, processing, and reconstructing features in the frequency domain, this module enables the model to learn high-frequency details and low-frequency contours with greater precision. Second, we introduce a large receptive field attention (LRFA) module in the encoder. Combining deep separable convolutions with multi-head attention, LRFA efficiently captures global contextual information at low computational cost. Finally, in the skip connections and decoding path, our weighted contextual fusion module (WCF) module dynamically generates channel attention weights for one feature stream to another, achieving efficient adaptive feature fusion. Simulation experiments on multiple public medical image segmentation datasets demonstrate that our MFR-UNet outperforms several existing mainstream methods in key metrics such as Dice coefficient and IoU, proving its effectiveness in enhancing segmentation accuracy and boundary clarity. Journal Article IET Systems Biology 20 1 Institution of Engineering and Technology (IET) 1751-8849 1751-8857 feature fusion; large receptive field; medical image segmentation; wavelet 1 1 2026 2026-01-01 10.1049/syb2.70049 COLLEGE NANME Mathematics and Computer Science School COLLEGE CODE MACS Swansea University SU Library paid the OA fee (TA Institutional Deal) Doctoral Training Centre at Swansea University (Grant Number: RS718); UKRI (Grant Number: EP/W020408/1); Hebei Province Education Science ”14th Five-Year Plan” (Grant Number: 2303065) 2026-01-07T17:14:57.2390185 2025-12-10T12:16:49.7341739 Faculty of Science and Engineering School of Mathematics and Computer Science - Computer Science Shaoqiang Wang 0000-0002-7539-5970 1 Guiling Shi 2 Shuo Sun 3 Yuchen Wang 4 Yulin Zhang 0000-0002-9125-5273 5 Weixian Li 6 Yawu Zhao 7 Cheng Cheng 0000-0003-0371-9646 8 71127__35915__d68c22c5d2884147b900afebd9d27ecc.pdf 71127.VoR.pdf 2026-01-07T17:12:49.2584562 Output 1770955 application/pdf Version of Record true © 2025 The Author(s). This is an open access article under the terms of the Creative Commons Attribution License. true eng http://creativecommons.org/licenses/by/4.0/
title MFR‐UNet: A Medical Image Segmentation Network With Fused Multi‐Scale Feature Refinement
spellingShingle MFR‐UNet: A Medical Image Segmentation Network With Fused Multi‐Scale Feature Refinement
Cheng Cheng
title_short MFR‐UNet: A Medical Image Segmentation Network With Fused Multi‐Scale Feature Refinement
title_full MFR‐UNet: A Medical Image Segmentation Network With Fused Multi‐Scale Feature Refinement
title_fullStr MFR‐UNet: A Medical Image Segmentation Network With Fused Multi‐Scale Feature Refinement
title_full_unstemmed MFR‐UNet: A Medical Image Segmentation Network With Fused Multi‐Scale Feature Refinement
title_sort MFR‐UNet: A Medical Image Segmentation Network With Fused Multi‐Scale Feature Refinement
author_id_str_mv 11ddf61c123b99e59b00fa1479367582
author_id_fullname_str_mv 11ddf61c123b99e59b00fa1479367582_***_Cheng Cheng
author Cheng Cheng
author2 Shaoqiang Wang
Guiling Shi
Shuo Sun
Yuchen Wang
Yulin Zhang
Weixian Li
Yawu Zhao
Cheng Cheng
format Journal article
container_title IET Systems Biology
container_volume 20
container_issue 1
publishDate 2026
institution Swansea University
issn 1751-8849
1751-8857
doi_str_mv 10.1049/syb2.70049
publisher Institution of Engineering and Technology (IET)
college_str Faculty of Science and Engineering
hierarchytype
hierarchy_top_id facultyofscienceandengineering
hierarchy_top_title Faculty of Science and Engineering
hierarchy_parent_id facultyofscienceandengineering
hierarchy_parent_title Faculty of Science and Engineering
department_str School of Mathematics and Computer Science - Computer Science{{{_:::_}}}Faculty of Science and Engineering{{{_:::_}}}School of Mathematics and Computer Science - Computer Science
document_store_str 1
active_str 0
description Medical image segmentation is crucial for clinical diagnosis and treatment planning. Although methods based on CNN, particularly U-Net and its variants, have achieved remarkable success in automated segmentation tasks, they still face challenges in effectively capturing long-range dependencies, refining multi-level features, and efficiently integrating cross-level information. To address these issues, we propose a novel U-Net architecture incorporating a multi-scale feature refinement mechanism (MFR-UNet). This network enhances segmentation accuracy and robustness by integrating three innovative modules. First, we designed a wavelet transform convolution (WtConv) module. By decomposing, processing, and reconstructing features in the frequency domain, this module enables the model to learn high-frequency details and low-frequency contours with greater precision. Second, we introduce a large receptive field attention (LRFA) module in the encoder. Combining deep separable convolutions with multi-head attention, LRFA efficiently captures global contextual information at low computational cost. Finally, in the skip connections and decoding path, our weighted contextual fusion module (WCF) module dynamically generates channel attention weights for one feature stream to another, achieving efficient adaptive feature fusion. Simulation experiments on multiple public medical image segmentation datasets demonstrate that our MFR-UNet outperforms several existing mainstream methods in key metrics such as Dice coefficient and IoU, proving its effectiveness in enhancing segmentation accuracy and boundary clarity.
published_date 2026-01-01T05:33:17Z
_version_ 1856805795592667136
score 11.09611

Similar Items