MCI Net: Mamba- Convolutional lightweight self-attention medical image segmentation network
With the development of deep learning in the field of medical image segmentation, various network segmentation models have been developed. Currently, the most common network models in medical image segmentation can be roughly categorized into pure convolutional networks, Transformer-based networks,...
Gespeichert in:
Veröffentlicht in: | Biomedical physics & engineering express 2024-11, Vol.11 (1), p.15003 |
---|---|
Hauptverfasser: | , , , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
container_end_page | |
---|---|
container_issue | 1 |
container_start_page | 15003 |
container_title | Biomedical physics & engineering express |
container_volume | 11 |
creator | Zhang, Yelin Wang, Guanglei Ma, Pengchong Li, Yan |
description | With the development of deep learning in the field of medical image segmentation, various network segmentation models have been developed. Currently, the most common network models in medical image segmentation can be roughly categorized into pure convolutional networks, Transformer-based networks, and networks combining convolution and Transformer architectures. However, when dealing with complex variations and irregular shapes in medical images, existing networks face issues such as incomplete information extraction, large model parameter sizes, high computational complexity, and long processing times. In contrast, models with lower parameter counts and complexity can efficiently, quickly, and accurately identify lesion areas, significantly reducing diagnosis time and providing valuable time for subsequent treatments. Therefore, this paper proposes a lightweight network named MCI-Net, with only 5.48 M parameters, a computational complexity of 4.41, and a time complexity of just 0.263. By performing linear modeling on sequences, MCI-Net permanently marks effective features and filters out irrelevant information. It efficiently captures local-global information with a small number of channels, reduces the number of parameters, and utilizes attention calculations with exchange value mapping. This achieves model lightweighting and enables thorough interaction of local-global information within the computation, establishing an overall semantic relationship of local-global information. To verify the effectiveness of the MCI-Net network, we conducted comparative experiments with other advanced representative networks on five public datasets: X-ray, Lung, ISIC-2016, ISIC-2018, and capsule endoscopy and gastrointestinal segmentation. We also performed ablation experiments on the first four datasets. The experimental results outperformed the other compared networks, confirming the effectiveness of MCI-Net. This research provides a valuable reference for achieving lightweight, accurate, and high-performance medical image segmentation network models. |
doi_str_mv | 10.1088/2057-1976/ad8acb |
format | Article |
fullrecord | <record><control><sourceid>proquest_iop_j</sourceid><recordid>TN_cdi_iop_journals_10_1088_2057_1976_ad8acb</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>3120596309</sourcerecordid><originalsourceid>FETCH-LOGICAL-c1701-a1ac46117667c96dd3b9509a7b827b4bcbf0cf6c3ce80a5639d4bb9aef6387393</originalsourceid><addsrcrecordid>eNp1kD1PwzAURS0Eoqh0Z0IZGQi148SO2VDFR6UWFpgYrGfHKSlJXOKEin-PQ0rFwmJbvudd6R2Ezgi-IjhNpxFOeEgEZ1PIUtDqAJ3svw7_vEdo4twaY0xYxJhIjtGIijjmiYhO0OtyNg8eTXsdLKFSEAYzW3_asmsLW0MZlMXqrd2a_gycKfMQ2tbUfRhUJiu0R4oKVsaHq8oH8BPVpt3a5v0UHeVQOjPZ3WP0cnf7PHsIF0_389nNItSEYxICAR0zQjhjXAuWZVSJBAvgKo24ipVWOdY501SbFEPCqMhipQSYnNGUU0HH6GLo3TT2ozOulVXhtClLqI3tnKTEuxCM4h7FA6ob61xjcrlp_ALNlyRY9lZlr0322uRg1Y-c79o75XfeD_w69MDlABR2I9e2a7w493_fN-RgghE</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>3120596309</pqid></control><display><type>article</type><title>MCI Net: Mamba- Convolutional lightweight self-attention medical image segmentation network</title><source>MEDLINE</source><source>Institute of Physics Journals</source><creator>Zhang, Yelin ; Wang, Guanglei ; Ma, Pengchong ; Li, Yan</creator><creatorcontrib>Zhang, Yelin ; Wang, Guanglei ; Ma, Pengchong ; Li, Yan</creatorcontrib><description>With the development of deep learning in the field of medical image segmentation, various network segmentation models have been developed. Currently, the most common network models in medical image segmentation can be roughly categorized into pure convolutional networks, Transformer-based networks, and networks combining convolution and Transformer architectures. However, when dealing with complex variations and irregular shapes in medical images, existing networks face issues such as incomplete information extraction, large model parameter sizes, high computational complexity, and long processing times. In contrast, models with lower parameter counts and complexity can efficiently, quickly, and accurately identify lesion areas, significantly reducing diagnosis time and providing valuable time for subsequent treatments. Therefore, this paper proposes a lightweight network named MCI-Net, with only 5.48 M parameters, a computational complexity of 4.41, and a time complexity of just 0.263. By performing linear modeling on sequences, MCI-Net permanently marks effective features and filters out irrelevant information. It efficiently captures local-global information with a small number of channels, reduces the number of parameters, and utilizes attention calculations with exchange value mapping. This achieves model lightweighting and enables thorough interaction of local-global information within the computation, establishing an overall semantic relationship of local-global information. To verify the effectiveness of the MCI-Net network, we conducted comparative experiments with other advanced representative networks on five public datasets: X-ray, Lung, ISIC-2016, ISIC-2018, and capsule endoscopy and gastrointestinal segmentation. We also performed ablation experiments on the first four datasets. The experimental results outperformed the other compared networks, confirming the effectiveness of MCI-Net. This research provides a valuable reference for achieving lightweight, accurate, and high-performance medical image segmentation network models.</description><identifier>ISSN: 2057-1976</identifier><identifier>EISSN: 2057-1976</identifier><identifier>DOI: 10.1088/2057-1976/ad8acb</identifier><identifier>PMID: 39447592</identifier><language>eng</language><publisher>England: IOP Publishing</publisher><subject>Algorithms ; attention computation ; Databases, Factual ; Deep Learning ; Diagnostic Imaging - methods ; Humans ; Image Processing, Computer-Assisted - methods ; mamba ; medical image segmentation ; Neural Networks, Computer ; Tomography, X-Ray Computed - methods ; transformer ; unet</subject><ispartof>Biomedical physics & engineering express, 2024-11, Vol.11 (1), p.15003</ispartof><rights>2024 IOP Publishing Ltd. All rights, including for text and data mining, AI training, and similar technologies, are reserved.</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><cites>FETCH-LOGICAL-c1701-a1ac46117667c96dd3b9509a7b827b4bcbf0cf6c3ce80a5639d4bb9aef6387393</cites><orcidid>0000-0002-2309-1510</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktopdf>$$Uhttps://iopscience.iop.org/article/10.1088/2057-1976/ad8acb/pdf$$EPDF$$P50$$Giop$$H</linktopdf><link.rule.ids>315,782,786,27931,27932,53853,53900</link.rule.ids><backlink>$$Uhttps://www.ncbi.nlm.nih.gov/pubmed/39447592$$D View this record in MEDLINE/PubMed$$Hfree_for_read</backlink></links><search><creatorcontrib>Zhang, Yelin</creatorcontrib><creatorcontrib>Wang, Guanglei</creatorcontrib><creatorcontrib>Ma, Pengchong</creatorcontrib><creatorcontrib>Li, Yan</creatorcontrib><title>MCI Net: Mamba- Convolutional lightweight self-attention medical image segmentation network</title><title>Biomedical physics & engineering express</title><addtitle>BPEX</addtitle><addtitle>Biomed. Phys. Eng. Express</addtitle><description>With the development of deep learning in the field of medical image segmentation, various network segmentation models have been developed. Currently, the most common network models in medical image segmentation can be roughly categorized into pure convolutional networks, Transformer-based networks, and networks combining convolution and Transformer architectures. However, when dealing with complex variations and irregular shapes in medical images, existing networks face issues such as incomplete information extraction, large model parameter sizes, high computational complexity, and long processing times. In contrast, models with lower parameter counts and complexity can efficiently, quickly, and accurately identify lesion areas, significantly reducing diagnosis time and providing valuable time for subsequent treatments. Therefore, this paper proposes a lightweight network named MCI-Net, with only 5.48 M parameters, a computational complexity of 4.41, and a time complexity of just 0.263. By performing linear modeling on sequences, MCI-Net permanently marks effective features and filters out irrelevant information. It efficiently captures local-global information with a small number of channels, reduces the number of parameters, and utilizes attention calculations with exchange value mapping. This achieves model lightweighting and enables thorough interaction of local-global information within the computation, establishing an overall semantic relationship of local-global information. To verify the effectiveness of the MCI-Net network, we conducted comparative experiments with other advanced representative networks on five public datasets: X-ray, Lung, ISIC-2016, ISIC-2018, and capsule endoscopy and gastrointestinal segmentation. We also performed ablation experiments on the first four datasets. The experimental results outperformed the other compared networks, confirming the effectiveness of MCI-Net. This research provides a valuable reference for achieving lightweight, accurate, and high-performance medical image segmentation network models.</description><subject>Algorithms</subject><subject>attention computation</subject><subject>Databases, Factual</subject><subject>Deep Learning</subject><subject>Diagnostic Imaging - methods</subject><subject>Humans</subject><subject>Image Processing, Computer-Assisted - methods</subject><subject>mamba</subject><subject>medical image segmentation</subject><subject>Neural Networks, Computer</subject><subject>Tomography, X-Ray Computed - methods</subject><subject>transformer</subject><subject>unet</subject><issn>2057-1976</issn><issn>2057-1976</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><sourceid>EIF</sourceid><recordid>eNp1kD1PwzAURS0Eoqh0Z0IZGQi148SO2VDFR6UWFpgYrGfHKSlJXOKEin-PQ0rFwmJbvudd6R2Ezgi-IjhNpxFOeEgEZ1PIUtDqAJ3svw7_vEdo4twaY0xYxJhIjtGIijjmiYhO0OtyNg8eTXsdLKFSEAYzW3_asmsLW0MZlMXqrd2a_gycKfMQ2tbUfRhUJiu0R4oKVsaHq8oH8BPVpt3a5v0UHeVQOjPZ3WP0cnf7PHsIF0_389nNItSEYxICAR0zQjhjXAuWZVSJBAvgKo24ipVWOdY501SbFEPCqMhipQSYnNGUU0HH6GLo3TT2ozOulVXhtClLqI3tnKTEuxCM4h7FA6ob61xjcrlp_ALNlyRY9lZlr0322uRg1Y-c79o75XfeD_w69MDlABR2I9e2a7w493_fN-RgghE</recordid><startdate>20241105</startdate><enddate>20241105</enddate><creator>Zhang, Yelin</creator><creator>Wang, Guanglei</creator><creator>Ma, Pengchong</creator><creator>Li, Yan</creator><general>IOP Publishing</general><scope>CGR</scope><scope>CUY</scope><scope>CVF</scope><scope>ECM</scope><scope>EIF</scope><scope>NPM</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>7X8</scope><orcidid>https://orcid.org/0000-0002-2309-1510</orcidid></search><sort><creationdate>20241105</creationdate><title>MCI Net: Mamba- Convolutional lightweight self-attention medical image segmentation network</title><author>Zhang, Yelin ; Wang, Guanglei ; Ma, Pengchong ; Li, Yan</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c1701-a1ac46117667c96dd3b9509a7b827b4bcbf0cf6c3ce80a5639d4bb9aef6387393</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Algorithms</topic><topic>attention computation</topic><topic>Databases, Factual</topic><topic>Deep Learning</topic><topic>Diagnostic Imaging - methods</topic><topic>Humans</topic><topic>Image Processing, Computer-Assisted - methods</topic><topic>mamba</topic><topic>medical image segmentation</topic><topic>Neural Networks, Computer</topic><topic>Tomography, X-Ray Computed - methods</topic><topic>transformer</topic><topic>unet</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Zhang, Yelin</creatorcontrib><creatorcontrib>Wang, Guanglei</creatorcontrib><creatorcontrib>Ma, Pengchong</creatorcontrib><creatorcontrib>Li, Yan</creatorcontrib><collection>Medline</collection><collection>MEDLINE</collection><collection>MEDLINE (Ovid)</collection><collection>MEDLINE</collection><collection>MEDLINE</collection><collection>PubMed</collection><collection>CrossRef</collection><collection>MEDLINE - Academic</collection><jtitle>Biomedical physics & engineering express</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Zhang, Yelin</au><au>Wang, Guanglei</au><au>Ma, Pengchong</au><au>Li, Yan</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>MCI Net: Mamba- Convolutional lightweight self-attention medical image segmentation network</atitle><jtitle>Biomedical physics & engineering express</jtitle><stitle>BPEX</stitle><addtitle>Biomed. Phys. Eng. Express</addtitle><date>2024-11-05</date><risdate>2024</risdate><volume>11</volume><issue>1</issue><spage>15003</spage><pages>15003-</pages><issn>2057-1976</issn><eissn>2057-1976</eissn><abstract>With the development of deep learning in the field of medical image segmentation, various network segmentation models have been developed. Currently, the most common network models in medical image segmentation can be roughly categorized into pure convolutional networks, Transformer-based networks, and networks combining convolution and Transformer architectures. However, when dealing with complex variations and irregular shapes in medical images, existing networks face issues such as incomplete information extraction, large model parameter sizes, high computational complexity, and long processing times. In contrast, models with lower parameter counts and complexity can efficiently, quickly, and accurately identify lesion areas, significantly reducing diagnosis time and providing valuable time for subsequent treatments. Therefore, this paper proposes a lightweight network named MCI-Net, with only 5.48 M parameters, a computational complexity of 4.41, and a time complexity of just 0.263. By performing linear modeling on sequences, MCI-Net permanently marks effective features and filters out irrelevant information. It efficiently captures local-global information with a small number of channels, reduces the number of parameters, and utilizes attention calculations with exchange value mapping. This achieves model lightweighting and enables thorough interaction of local-global information within the computation, establishing an overall semantic relationship of local-global information. To verify the effectiveness of the MCI-Net network, we conducted comparative experiments with other advanced representative networks on five public datasets: X-ray, Lung, ISIC-2016, ISIC-2018, and capsule endoscopy and gastrointestinal segmentation. We also performed ablation experiments on the first four datasets. The experimental results outperformed the other compared networks, confirming the effectiveness of MCI-Net. This research provides a valuable reference for achieving lightweight, accurate, and high-performance medical image segmentation network models.</abstract><cop>England</cop><pub>IOP Publishing</pub><pmid>39447592</pmid><doi>10.1088/2057-1976/ad8acb</doi><tpages>15</tpages><orcidid>https://orcid.org/0000-0002-2309-1510</orcidid></addata></record> |
fulltext | fulltext |
identifier | ISSN: 2057-1976 |
ispartof | Biomedical physics & engineering express, 2024-11, Vol.11 (1), p.15003 |
issn | 2057-1976 2057-1976 |
language | eng |
recordid | cdi_iop_journals_10_1088_2057_1976_ad8acb |
source | MEDLINE; Institute of Physics Journals |
subjects | Algorithms attention computation Databases, Factual Deep Learning Diagnostic Imaging - methods Humans Image Processing, Computer-Assisted - methods mamba medical image segmentation Neural Networks, Computer Tomography, X-Ray Computed - methods transformer unet |
title | MCI Net: Mamba- Convolutional lightweight self-attention medical image segmentation network |
url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-06T07%3A25%3A27IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_iop_j&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=MCI%20Net:%20Mamba-%20Convolutional%20lightweight%20self-attention%20medical%20image%20segmentation%20network&rft.jtitle=Biomedical%20physics%20&%20engineering%20express&rft.au=Zhang,%20Yelin&rft.date=2024-11-05&rft.volume=11&rft.issue=1&rft.spage=15003&rft.pages=15003-&rft.issn=2057-1976&rft.eissn=2057-1976&rft_id=info:doi/10.1088/2057-1976/ad8acb&rft_dat=%3Cproquest_iop_j%3E3120596309%3C/proquest_iop_j%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=3120596309&rft_id=info:pmid/39447592&rfr_iscdi=true |