Multi-planar dual adversarial network based on dynamic 3D features for MRI-CT head and neck image synthesis

Head and neck radiotherapy planning requires electron densities from different tissues for dose calculation. Dose calculation from imaging modalities such as MRI remains an unsolved problem since this imaging modality does not provide information about the density of electrons. Approach. We propose...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Physics in medicine & biology 2024-08, Vol.69 (15), p.155012
Hauptverfasser: Touati, Redha, Trung Le, William, Kadoury, Samuel
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue 15
container_start_page 155012
container_title Physics in medicine & biology
container_volume 69
creator Touati, Redha
Trung Le, William
Kadoury, Samuel
description Head and neck radiotherapy planning requires electron densities from different tissues for dose calculation. Dose calculation from imaging modalities such as MRI remains an unsolved problem since this imaging modality does not provide information about the density of electrons. Approach. We propose a generative adversarial network (GAN) approach that synthesizes CT (sCT) images from T1-weighted MRI acquisitions in head and neck cancer patients. Our contribution is to exploit new features that are relevant for improving multimodal image synthesis, and thus improving the quality of the generated CT images. More precisely, we propose a Dual branch generator based on the U-Net architecture and on an augmented multi-planar branch. The augmented branch learns specific 3D dynamic features, which describe the dynamic image shape variations and are extracted from different view-points of the volumetric input MRI. The architecture of the proposed model relies on an end-to-end convolutional U-Net embedding network. Results. The proposed model achieves a mean absolute error (MAE) of $18.76 (5.167)$ in the target Hounsfield unit (HU) space on sagittal head and neck patients, with a mean structural similarity (MSSIM) of $0.95 (0.09)$ and a Frechet inception distance (FID) of $145.60 (8.38)$. The model yields a MAE of $26.83 (8.27)$ to generate specific primary tumor regions on axial patient acquisitions, with a Dice score of $0.73 (0.06)$ and a FID distance equal to $122.58 (7.55)$. The improvement of our model over other state-of-the-art GAN approaches is of $3.8\%$, on a tumor test set. On both sagittal and axial acquisitions, the model yields the best peak signal-to-noise ratio (PSNRs) of $27.89 (2.22)$ and $26.08 (2.95)$ to synthesize MRI from CT input. Significance. The proposed model synthesizes both sagittal and axial CT tumor images, used for radiotherapy treatment planning in head and neck cancer cases. The performance analysis across different imaging metrics and under different evaluation strategies demonstrates the effectiveness of our dual CT synthesis model to produce high quality sCT images compared to other state-of-the-art approaches. Our model could improve clinical tumor analysis, in which a further clinical validation remains to be explored.
doi_str_mv 10.1088/1361-6560/ad611a
format Article
fullrecord <record><control><sourceid>proquest_iop_j</sourceid><recordid>TN_cdi_iop_journals_10_1088_1361_6560_ad611a</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>3077990892</sourcerecordid><originalsourceid>FETCH-LOGICAL-c294t-a4d3e4be678674cd063c04a6c437b0ea660dbd3d118a7e6cdfe75d94dd174e3</originalsourceid><addsrcrecordid>eNp9kM1P3DAQxa2qqCy0954qHzk0xV47TnysthSQQEiwd2vimZSw-aqdFO1_X6924YSQRvLI-r2neY-xr1L8kKIsz6UyMjO5EeeARkr4wBavXx_ZQgglMyvz_JidxPgkhJTlUn9ix6q0pcytWrDN7dxOTTa20EPgOEPLAf9RiBCatPc0PQ9hwyuIhHzoOW576BrP1S9eE0xzoMjrIfDb--tsteaPBMihxyT0G9508Id43PbTI8UmfmZHNbSRvhzeU_bw-2K9uspu7i6vVz9vMr-0espAoyJdkSlKU2iPwigvNBivVVEJAmMEVqgwhYGCjMeaihytRpSFJnXKzvauYxj-zhQn1zXRU5sS0jBHp0RRWCtKu0yo2KM-DDEGqt0Y0s1h66Rwu4Ldrk23a9PtC06Sbwf3ueoIXwUvjSbg-x5ohtE9DXPoU9T3_M7ewMeucsY6mafJhVy6EWv1HyYkkkM</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>3077990892</pqid></control><display><type>article</type><title>Multi-planar dual adversarial network based on dynamic 3D features for MRI-CT head and neck image synthesis</title><source>HEAL-Link subscriptions: Institute of Physics (IOP) Journals</source><source>Institute of Physics Journals</source><creator>Touati, Redha ; Trung Le, William ; Kadoury, Samuel</creator><creatorcontrib>Touati, Redha ; Trung Le, William ; Kadoury, Samuel</creatorcontrib><description>Head and neck radiotherapy planning requires electron densities from different tissues for dose calculation. Dose calculation from imaging modalities such as MRI remains an unsolved problem since this imaging modality does not provide information about the density of electrons. Approach. We propose a generative adversarial network (GAN) approach that synthesizes CT (sCT) images from T1-weighted MRI acquisitions in head and neck cancer patients. Our contribution is to exploit new features that are relevant for improving multimodal image synthesis, and thus improving the quality of the generated CT images. More precisely, we propose a Dual branch generator based on the U-Net architecture and on an augmented multi-planar branch. The augmented branch learns specific 3D dynamic features, which describe the dynamic image shape variations and are extracted from different view-points of the volumetric input MRI. The architecture of the proposed model relies on an end-to-end convolutional U-Net embedding network. Results. The proposed model achieves a mean absolute error (MAE) of $18.76 (5.167)$ in the target Hounsfield unit (HU) space on sagittal head and neck patients, with a mean structural similarity (MSSIM) of $0.95 (0.09)$ and a Frechet inception distance (FID) of $145.60 (8.38)$. The model yields a MAE of $26.83 (8.27)$ to generate specific primary tumor regions on axial patient acquisitions, with a Dice score of $0.73 (0.06)$ and a FID distance equal to $122.58 (7.55)$. The improvement of our model over other state-of-the-art GAN approaches is of $3.8\%$, on a tumor test set. On both sagittal and axial acquisitions, the model yields the best peak signal-to-noise ratio (PSNRs) of $27.89 (2.22)$ and $26.08 (2.95)$ to synthesize MRI from CT input. Significance. The proposed model synthesizes both sagittal and axial CT tumor images, used for radiotherapy treatment planning in head and neck cancer cases. The performance analysis across different imaging metrics and under different evaluation strategies demonstrates the effectiveness of our dual CT synthesis model to produce high quality sCT images compared to other state-of-the-art approaches. Our model could improve clinical tumor analysis, in which a further clinical validation remains to be explored.</description><identifier>ISSN: 0031-9155</identifier><identifier>ISSN: 1361-6560</identifier><identifier>EISSN: 1361-6560</identifier><identifier>DOI: 10.1088/1361-6560/ad611a</identifier><identifier>PMID: 38981593</identifier><identifier>CODEN: PHMBA7</identifier><language>eng</language><publisher>England: IOP Publishing</publisher><subject>3D multi-view image modeling ; adversarial network ; dual feature learning ; dynamic features ; generative network model ; image generation</subject><ispartof>Physics in medicine &amp; biology, 2024-08, Vol.69 (15), p.155012</ispartof><rights>2024 The Author(s). Published on behalf of Institute of Physics and Engineering in Medicine by IOP Publishing Ltd</rights><rights>Creative Commons Attribution license.</rights><lds50>peer_reviewed</lds50><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed><cites>FETCH-LOGICAL-c294t-a4d3e4be678674cd063c04a6c437b0ea660dbd3d118a7e6cdfe75d94dd174e3</cites><orcidid>0000-0003-3845-5361 ; 0000-0002-3048-4291</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktopdf>$$Uhttps://iopscience.iop.org/article/10.1088/1361-6560/ad611a/pdf$$EPDF$$P50$$Giop$$Hfree_for_read</linktopdf><link.rule.ids>314,780,784,27924,27925,53846,53893</link.rule.ids><backlink>$$Uhttps://www.ncbi.nlm.nih.gov/pubmed/38981593$$D View this record in MEDLINE/PubMed$$Hfree_for_read</backlink></links><search><creatorcontrib>Touati, Redha</creatorcontrib><creatorcontrib>Trung Le, William</creatorcontrib><creatorcontrib>Kadoury, Samuel</creatorcontrib><title>Multi-planar dual adversarial network based on dynamic 3D features for MRI-CT head and neck image synthesis</title><title>Physics in medicine &amp; biology</title><addtitle>PMB</addtitle><addtitle>Phys. Med. Biol</addtitle><description>Head and neck radiotherapy planning requires electron densities from different tissues for dose calculation. Dose calculation from imaging modalities such as MRI remains an unsolved problem since this imaging modality does not provide information about the density of electrons. Approach. We propose a generative adversarial network (GAN) approach that synthesizes CT (sCT) images from T1-weighted MRI acquisitions in head and neck cancer patients. Our contribution is to exploit new features that are relevant for improving multimodal image synthesis, and thus improving the quality of the generated CT images. More precisely, we propose a Dual branch generator based on the U-Net architecture and on an augmented multi-planar branch. The augmented branch learns specific 3D dynamic features, which describe the dynamic image shape variations and are extracted from different view-points of the volumetric input MRI. The architecture of the proposed model relies on an end-to-end convolutional U-Net embedding network. Results. The proposed model achieves a mean absolute error (MAE) of $18.76 (5.167)$ in the target Hounsfield unit (HU) space on sagittal head and neck patients, with a mean structural similarity (MSSIM) of $0.95 (0.09)$ and a Frechet inception distance (FID) of $145.60 (8.38)$. The model yields a MAE of $26.83 (8.27)$ to generate specific primary tumor regions on axial patient acquisitions, with a Dice score of $0.73 (0.06)$ and a FID distance equal to $122.58 (7.55)$. The improvement of our model over other state-of-the-art GAN approaches is of $3.8\%$, on a tumor test set. On both sagittal and axial acquisitions, the model yields the best peak signal-to-noise ratio (PSNRs) of $27.89 (2.22)$ and $26.08 (2.95)$ to synthesize MRI from CT input. Significance. The proposed model synthesizes both sagittal and axial CT tumor images, used for radiotherapy treatment planning in head and neck cancer cases. The performance analysis across different imaging metrics and under different evaluation strategies demonstrates the effectiveness of our dual CT synthesis model to produce high quality sCT images compared to other state-of-the-art approaches. Our model could improve clinical tumor analysis, in which a further clinical validation remains to be explored.</description><subject>3D multi-view image modeling</subject><subject>adversarial network</subject><subject>dual feature learning</subject><subject>dynamic features</subject><subject>generative network model</subject><subject>image generation</subject><issn>0031-9155</issn><issn>1361-6560</issn><issn>1361-6560</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><sourceid>O3W</sourceid><recordid>eNp9kM1P3DAQxa2qqCy0954qHzk0xV47TnysthSQQEiwd2vimZSw-aqdFO1_X6924YSQRvLI-r2neY-xr1L8kKIsz6UyMjO5EeeARkr4wBavXx_ZQgglMyvz_JidxPgkhJTlUn9ix6q0pcytWrDN7dxOTTa20EPgOEPLAf9RiBCatPc0PQ9hwyuIhHzoOW576BrP1S9eE0xzoMjrIfDb--tsteaPBMihxyT0G9508Id43PbTI8UmfmZHNbSRvhzeU_bw-2K9uspu7i6vVz9vMr-0espAoyJdkSlKU2iPwigvNBivVVEJAmMEVqgwhYGCjMeaihytRpSFJnXKzvauYxj-zhQn1zXRU5sS0jBHp0RRWCtKu0yo2KM-DDEGqt0Y0s1h66Rwu4Ldrk23a9PtC06Sbwf3ueoIXwUvjSbg-x5ohtE9DXPoU9T3_M7ewMeucsY6mafJhVy6EWv1HyYkkkM</recordid><startdate>20240807</startdate><enddate>20240807</enddate><creator>Touati, Redha</creator><creator>Trung Le, William</creator><creator>Kadoury, Samuel</creator><general>IOP Publishing</general><scope>O3W</scope><scope>TSCCA</scope><scope>NPM</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>7X8</scope><orcidid>https://orcid.org/0000-0003-3845-5361</orcidid><orcidid>https://orcid.org/0000-0002-3048-4291</orcidid></search><sort><creationdate>20240807</creationdate><title>Multi-planar dual adversarial network based on dynamic 3D features for MRI-CT head and neck image synthesis</title><author>Touati, Redha ; Trung Le, William ; Kadoury, Samuel</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c294t-a4d3e4be678674cd063c04a6c437b0ea660dbd3d118a7e6cdfe75d94dd174e3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>3D multi-view image modeling</topic><topic>adversarial network</topic><topic>dual feature learning</topic><topic>dynamic features</topic><topic>generative network model</topic><topic>image generation</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Touati, Redha</creatorcontrib><creatorcontrib>Trung Le, William</creatorcontrib><creatorcontrib>Kadoury, Samuel</creatorcontrib><collection>Open Access: IOP Publishing Free Content</collection><collection>IOPscience (Open Access)</collection><collection>PubMed</collection><collection>CrossRef</collection><collection>MEDLINE - Academic</collection><jtitle>Physics in medicine &amp; biology</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Touati, Redha</au><au>Trung Le, William</au><au>Kadoury, Samuel</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Multi-planar dual adversarial network based on dynamic 3D features for MRI-CT head and neck image synthesis</atitle><jtitle>Physics in medicine &amp; biology</jtitle><stitle>PMB</stitle><addtitle>Phys. Med. Biol</addtitle><date>2024-08-07</date><risdate>2024</risdate><volume>69</volume><issue>15</issue><spage>155012</spage><pages>155012-</pages><issn>0031-9155</issn><issn>1361-6560</issn><eissn>1361-6560</eissn><coden>PHMBA7</coden><abstract>Head and neck radiotherapy planning requires electron densities from different tissues for dose calculation. Dose calculation from imaging modalities such as MRI remains an unsolved problem since this imaging modality does not provide information about the density of electrons. Approach. We propose a generative adversarial network (GAN) approach that synthesizes CT (sCT) images from T1-weighted MRI acquisitions in head and neck cancer patients. Our contribution is to exploit new features that are relevant for improving multimodal image synthesis, and thus improving the quality of the generated CT images. More precisely, we propose a Dual branch generator based on the U-Net architecture and on an augmented multi-planar branch. The augmented branch learns specific 3D dynamic features, which describe the dynamic image shape variations and are extracted from different view-points of the volumetric input MRI. The architecture of the proposed model relies on an end-to-end convolutional U-Net embedding network. Results. The proposed model achieves a mean absolute error (MAE) of $18.76 (5.167)$ in the target Hounsfield unit (HU) space on sagittal head and neck patients, with a mean structural similarity (MSSIM) of $0.95 (0.09)$ and a Frechet inception distance (FID) of $145.60 (8.38)$. The model yields a MAE of $26.83 (8.27)$ to generate specific primary tumor regions on axial patient acquisitions, with a Dice score of $0.73 (0.06)$ and a FID distance equal to $122.58 (7.55)$. The improvement of our model over other state-of-the-art GAN approaches is of $3.8\%$, on a tumor test set. On both sagittal and axial acquisitions, the model yields the best peak signal-to-noise ratio (PSNRs) of $27.89 (2.22)$ and $26.08 (2.95)$ to synthesize MRI from CT input. Significance. The proposed model synthesizes both sagittal and axial CT tumor images, used for radiotherapy treatment planning in head and neck cancer cases. The performance analysis across different imaging metrics and under different evaluation strategies demonstrates the effectiveness of our dual CT synthesis model to produce high quality sCT images compared to other state-of-the-art approaches. Our model could improve clinical tumor analysis, in which a further clinical validation remains to be explored.</abstract><cop>England</cop><pub>IOP Publishing</pub><pmid>38981593</pmid><doi>10.1088/1361-6560/ad611a</doi><tpages>19</tpages><orcidid>https://orcid.org/0000-0003-3845-5361</orcidid><orcidid>https://orcid.org/0000-0002-3048-4291</orcidid><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier ISSN: 0031-9155
ispartof Physics in medicine & biology, 2024-08, Vol.69 (15), p.155012
issn 0031-9155
1361-6560
1361-6560
language eng
recordid cdi_iop_journals_10_1088_1361_6560_ad611a
source HEAL-Link subscriptions: Institute of Physics (IOP) Journals; Institute of Physics Journals
subjects 3D multi-view image modeling
adversarial network
dual feature learning
dynamic features
generative network model
image generation
title Multi-planar dual adversarial network based on dynamic 3D features for MRI-CT head and neck image synthesis
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-25T06%3A24%3A34IST&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=Multi-planar%20dual%20adversarial%20network%20based%20on%20dynamic%203D%20features%20for%20MRI-CT%20head%20and%20neck%20image%20synthesis&rft.jtitle=Physics%20in%20medicine%20&%20biology&rft.au=Touati,%20Redha&rft.date=2024-08-07&rft.volume=69&rft.issue=15&rft.spage=155012&rft.pages=155012-&rft.issn=0031-9155&rft.eissn=1361-6560&rft.coden=PHMBA7&rft_id=info:doi/10.1088/1361-6560/ad611a&rft_dat=%3Cproquest_iop_j%3E3077990892%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=3077990892&rft_id=info:pmid/38981593&rfr_iscdi=true