Multimodal medical image‐to‐image translation via variational autoencoder latent space mapping
Background Medical image translation has become an essential tool in modern radiotherapy, providing complementary information for target delineation and dose calculation. However, current approaches are constrained by their modality‐specific nature, requiring separate model training for each pair of...
Uložené v:
| Vydané v: | Medical physics (Lancaster) Ročník 52; číslo 7; s. e17912 - n/a |
|---|---|
| Hlavní autori: | , , , , , |
| Médium: | Journal Article |
| Jazyk: | English |
| Vydavateľské údaje: |
United States
01.07.2025
|
| Predmet: | |
| ISSN: | 0094-2405, 2473-4209, 2473-4209 |
| On-line prístup: | Získať plný text |
| Tagy: |
Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
|
| Abstract | Background
Medical image translation has become an essential tool in modern radiotherapy, providing complementary information for target delineation and dose calculation. However, current approaches are constrained by their modality‐specific nature, requiring separate model training for each pair of imaging modalities. This limitation hinders the efficient deployment of comprehensive multimodal solutions in clinical practice.
Purpose
To develop a unified image translation method using variational autoencoder (VAE) latent space mapping, which enables flexible conversion between different medical imaging modalities to meet clinical demands.
Methods
We propose a three‐stage approach to construct a unified image translation model. Initially, a VAE is trained to learn a shared latent space for various medical images. A stacked bidirectional transformer is subsequently utilized to learn the mapping between different modalities within the latent space under the guidance of the image modality. Finally, the VAE decoder is fine‐tuned to improve image quality. Our internal dataset collected paired imaging data from 87 head and neck cases, with each case containing cone beam computed tomography (CBCT), computed tomography (CT), MR T1c, and MR T2W images. The effectiveness of this strategy is quantitatively evaluated on our internal dataset and a public dataset by the mean absolute error (MAE), peak‐signal‐to‐noise ratio (PSNR), and structural similarity index (SSIM). Additionally, the dosimetry characteristics of the synthetic CT images are evaluated, and subjective quality assessments of the synthetic MR images are conducted to determine their clinical value.
Results
The VAE with the Kullback‒Leibler (KL)‐16 image tokenizer demonstrates superior image reconstruction ability, achieving a Fréchet inception distance (FID) of 4.84, a PSNR of 32.80 dB, and an SSIM of 92.33%. In synthetic CT tasks, the model shows greater accuracy in intramodality translations than in cross‐modality translations, as evidenced by an MAE of 21.60 ± 8.80 Hounsfield unit (HU) in the CBCT‐to‐CT task and 45.23 ± 13.21 HU/47.55 ± 13.88 in the MR T1c/T2w‐to‐CT tasks. For the cross‐contrast MR translation tasks, the results are very close, with mean PSNR and SSIM values of 26.33 ± 1.36 dB and 85.21% ± 2.21%, respectively, for the T1c‐to‐T2w translation and 26.03 ± 1.67 dB and 85.73% ± 2.66%, respectively, for the T2w‐to‐T1c translation. Dosimetric results indicate that all the gamma pass rates for synthetic CTs are higher than 99% for photon intensity‐modulated radiation therapy (IMRT) planning. However, the subjective quality assessment scores for synthetic MR images are lower than those for real MR images.
Conclusions
The proposed three‐stage approach successfully develops a unified image translation model that can effectively handle a wide range of medical image translation tasks. This flexibility and effectiveness make it a valuable tool for clinical applications. |
|---|---|
| AbstractList | Medical image translation has become an essential tool in modern radiotherapy, providing complementary information for target delineation and dose calculation. However, current approaches are constrained by their modality-specific nature, requiring separate model training for each pair of imaging modalities. This limitation hinders the efficient deployment of comprehensive multimodal solutions in clinical practice.
To develop a unified image translation method using variational autoencoder (VAE) latent space mapping, which enables flexible conversion between different medical imaging modalities to meet clinical demands.
We propose a three-stage approach to construct a unified image translation model. Initially, a VAE is trained to learn a shared latent space for various medical images. A stacked bidirectional transformer is subsequently utilized to learn the mapping between different modalities within the latent space under the guidance of the image modality. Finally, the VAE decoder is fine-tuned to improve image quality. Our internal dataset collected paired imaging data from 87 head and neck cases, with each case containing cone beam computed tomography (CBCT), computed tomography (CT), MR T1c, and MR T2W images. The effectiveness of this strategy is quantitatively evaluated on our internal dataset and a public dataset by the mean absolute error (MAE), peak-signal-to-noise ratio (PSNR), and structural similarity index (SSIM). Additionally, the dosimetry characteristics of the synthetic CT images are evaluated, and subjective quality assessments of the synthetic MR images are conducted to determine their clinical value.
The VAE with the Kullback‒Leibler (KL)-16 image tokenizer demonstrates superior image reconstruction ability, achieving a Fréchet inception distance (FID) of 4.84, a PSNR of 32.80 dB, and an SSIM of 92.33%. In synthetic CT tasks, the model shows greater accuracy in intramodality translations than in cross-modality translations, as evidenced by an MAE of 21.60 ± 8.80 Hounsfield unit (HU) in the CBCT-to-CT task and 45.23 ± 13.21 HU/47.55 ± 13.88 in the MR T1c/T2w-to-CT tasks. For the cross-contrast MR translation tasks, the results are very close, with mean PSNR and SSIM values of 26.33 ± 1.36 dB and 85.21% ± 2.21%, respectively, for the T1c-to-T2w translation and 26.03 ± 1.67 dB and 85.73% ± 2.66%, respectively, for the T2w-to-T1c translation. Dosimetric results indicate that all the gamma pass rates for synthetic CTs are higher than 99% for photon intensity-modulated radiation therapy (IMRT) planning. However, the subjective quality assessment scores for synthetic MR images are lower than those for real MR images.
The proposed three-stage approach successfully develops a unified image translation model that can effectively handle a wide range of medical image translation tasks. This flexibility and effectiveness make it a valuable tool for clinical applications. Background Medical image translation has become an essential tool in modern radiotherapy, providing complementary information for target delineation and dose calculation. However, current approaches are constrained by their modality‐specific nature, requiring separate model training for each pair of imaging modalities. This limitation hinders the efficient deployment of comprehensive multimodal solutions in clinical practice. Purpose To develop a unified image translation method using variational autoencoder (VAE) latent space mapping, which enables flexible conversion between different medical imaging modalities to meet clinical demands. Methods We propose a three‐stage approach to construct a unified image translation model. Initially, a VAE is trained to learn a shared latent space for various medical images. A stacked bidirectional transformer is subsequently utilized to learn the mapping between different modalities within the latent space under the guidance of the image modality. Finally, the VAE decoder is fine‐tuned to improve image quality. Our internal dataset collected paired imaging data from 87 head and neck cases, with each case containing cone beam computed tomography (CBCT), computed tomography (CT), MR T1c, and MR T2W images. The effectiveness of this strategy is quantitatively evaluated on our internal dataset and a public dataset by the mean absolute error (MAE), peak‐signal‐to‐noise ratio (PSNR), and structural similarity index (SSIM). Additionally, the dosimetry characteristics of the synthetic CT images are evaluated, and subjective quality assessments of the synthetic MR images are conducted to determine their clinical value. Results The VAE with the Kullback‒Leibler (KL)‐16 image tokenizer demonstrates superior image reconstruction ability, achieving a Fréchet inception distance (FID) of 4.84, a PSNR of 32.80 dB, and an SSIM of 92.33%. In synthetic CT tasks, the model shows greater accuracy in intramodality translations than in cross‐modality translations, as evidenced by an MAE of 21.60 ± 8.80 Hounsfield unit (HU) in the CBCT‐to‐CT task and 45.23 ± 13.21 HU/47.55 ± 13.88 in the MR T1c/T2w‐to‐CT tasks. For the cross‐contrast MR translation tasks, the results are very close, with mean PSNR and SSIM values of 26.33 ± 1.36 dB and 85.21% ± 2.21%, respectively, for the T1c‐to‐T2w translation and 26.03 ± 1.67 dB and 85.73% ± 2.66%, respectively, for the T2w‐to‐T1c translation. Dosimetric results indicate that all the gamma pass rates for synthetic CTs are higher than 99% for photon intensity‐modulated radiation therapy (IMRT) planning. However, the subjective quality assessment scores for synthetic MR images are lower than those for real MR images. Conclusions The proposed three‐stage approach successfully develops a unified image translation model that can effectively handle a wide range of medical image translation tasks. This flexibility and effectiveness make it a valuable tool for clinical applications. Medical image translation has become an essential tool in modern radiotherapy, providing complementary information for target delineation and dose calculation. However, current approaches are constrained by their modality-specific nature, requiring separate model training for each pair of imaging modalities. This limitation hinders the efficient deployment of comprehensive multimodal solutions in clinical practice.BACKGROUNDMedical image translation has become an essential tool in modern radiotherapy, providing complementary information for target delineation and dose calculation. However, current approaches are constrained by their modality-specific nature, requiring separate model training for each pair of imaging modalities. This limitation hinders the efficient deployment of comprehensive multimodal solutions in clinical practice.To develop a unified image translation method using variational autoencoder (VAE) latent space mapping, which enables flexible conversion between different medical imaging modalities to meet clinical demands.PURPOSETo develop a unified image translation method using variational autoencoder (VAE) latent space mapping, which enables flexible conversion between different medical imaging modalities to meet clinical demands.We propose a three-stage approach to construct a unified image translation model. Initially, a VAE is trained to learn a shared latent space for various medical images. A stacked bidirectional transformer is subsequently utilized to learn the mapping between different modalities within the latent space under the guidance of the image modality. Finally, the VAE decoder is fine-tuned to improve image quality. Our internal dataset collected paired imaging data from 87 head and neck cases, with each case containing cone beam computed tomography (CBCT), computed tomography (CT), MR T1c, and MR T2W images. The effectiveness of this strategy is quantitatively evaluated on our internal dataset and a public dataset by the mean absolute error (MAE), peak-signal-to-noise ratio (PSNR), and structural similarity index (SSIM). Additionally, the dosimetry characteristics of the synthetic CT images are evaluated, and subjective quality assessments of the synthetic MR images are conducted to determine their clinical value.METHODSWe propose a three-stage approach to construct a unified image translation model. Initially, a VAE is trained to learn a shared latent space for various medical images. A stacked bidirectional transformer is subsequently utilized to learn the mapping between different modalities within the latent space under the guidance of the image modality. Finally, the VAE decoder is fine-tuned to improve image quality. Our internal dataset collected paired imaging data from 87 head and neck cases, with each case containing cone beam computed tomography (CBCT), computed tomography (CT), MR T1c, and MR T2W images. The effectiveness of this strategy is quantitatively evaluated on our internal dataset and a public dataset by the mean absolute error (MAE), peak-signal-to-noise ratio (PSNR), and structural similarity index (SSIM). Additionally, the dosimetry characteristics of the synthetic CT images are evaluated, and subjective quality assessments of the synthetic MR images are conducted to determine their clinical value.The VAE with the Kullback‒Leibler (KL)-16 image tokenizer demonstrates superior image reconstruction ability, achieving a Fréchet inception distance (FID) of 4.84, a PSNR of 32.80 dB, and an SSIM of 92.33%. In synthetic CT tasks, the model shows greater accuracy in intramodality translations than in cross-modality translations, as evidenced by an MAE of 21.60 ± 8.80 Hounsfield unit (HU) in the CBCT-to-CT task and 45.23 ± 13.21 HU/47.55 ± 13.88 in the MR T1c/T2w-to-CT tasks. For the cross-contrast MR translation tasks, the results are very close, with mean PSNR and SSIM values of 26.33 ± 1.36 dB and 85.21% ± 2.21%, respectively, for the T1c-to-T2w translation and 26.03 ± 1.67 dB and 85.73% ± 2.66%, respectively, for the T2w-to-T1c translation. Dosimetric results indicate that all the gamma pass rates for synthetic CTs are higher than 99% for photon intensity-modulated radiation therapy (IMRT) planning. However, the subjective quality assessment scores for synthetic MR images are lower than those for real MR images.RESULTSThe VAE with the Kullback‒Leibler (KL)-16 image tokenizer demonstrates superior image reconstruction ability, achieving a Fréchet inception distance (FID) of 4.84, a PSNR of 32.80 dB, and an SSIM of 92.33%. In synthetic CT tasks, the model shows greater accuracy in intramodality translations than in cross-modality translations, as evidenced by an MAE of 21.60 ± 8.80 Hounsfield unit (HU) in the CBCT-to-CT task and 45.23 ± 13.21 HU/47.55 ± 13.88 in the MR T1c/T2w-to-CT tasks. For the cross-contrast MR translation tasks, the results are very close, with mean PSNR and SSIM values of 26.33 ± 1.36 dB and 85.21% ± 2.21%, respectively, for the T1c-to-T2w translation and 26.03 ± 1.67 dB and 85.73% ± 2.66%, respectively, for the T2w-to-T1c translation. Dosimetric results indicate that all the gamma pass rates for synthetic CTs are higher than 99% for photon intensity-modulated radiation therapy (IMRT) planning. However, the subjective quality assessment scores for synthetic MR images are lower than those for real MR images.The proposed three-stage approach successfully develops a unified image translation model that can effectively handle a wide range of medical image translation tasks. This flexibility and effectiveness make it a valuable tool for clinical applications.CONCLUSIONSThe proposed three-stage approach successfully develops a unified image translation model that can effectively handle a wide range of medical image translation tasks. This flexibility and effectiveness make it a valuable tool for clinical applications. |
| Author | Li, Song Liang, Zhiwen Tian, Xin Hu, Ying Cheng, Mengjie Ma, Jinhui |
| Author_xml | – sequence: 1 givenname: Zhiwen surname: Liang fullname: Liang, Zhiwen organization: Hubei Key Laboratory of Precision Radiation Oncology – sequence: 2 givenname: Mengjie surname: Cheng fullname: Cheng, Mengjie organization: Renmin Hospital of Wuhan University – sequence: 3 givenname: Jinhui surname: Ma fullname: Ma, Jinhui organization: Union Hospital, Tongji Medical College, Huazhong University of Science and Technology – sequence: 4 givenname: Ying surname: Hu fullname: Hu, Ying organization: Hubei University of Education – sequence: 5 givenname: Song surname: Li fullname: Li, Song email: ls@whu.edu.cn organization: Wuhan University – sequence: 6 givenname: Xin surname: Tian fullname: Tian, Xin email: xin.tian@whu.edu.cn organization: Wuhan University |
| BackLink | https://www.ncbi.nlm.nih.gov/pubmed/40439703$$D View this record in MEDLINE/PubMed |
| BookMark | eNp10MlOwzAQBmALFdEFJJ4A5cglZbxkO6KKTWoFBzhbjuNURokd7KSoNx6BZ-RJMC3LicuMRvo0mvmnaGSsUQidYphjAHLRdnOcFZgcoAlhGY0ZgWKEJgAFiwmDZIym3j8DQEoTOEJjBowWGdAJKldD0-vWVqKJWlVpGbpuxVp9vL33NpTdEPVOGN-IXlsTbbSINsLp3RS4GHqrjLSVclEgyvSR74RUUSu6Tpv1MTqsRePVyXefoafrq8fFbby8v7lbXC5jSXJG4jSRCUtBUlnLnJUlwxiEEjUlVJV5xkpW4brOSxXuplnCCgEyx1meyJoWAgOdofP93s7Zl0H5nrfaS9U0wig7eE4JpikmJC8CPfumQxm-5p0Lb7ot_8nlb5d01nun6l-CgX9FztuO7yIPNN7TV92o7b-Orx72_hMeTYOY |
| Cites_doi | 10.1007/s44163‐021‐00006‐0 10.1007/978-3-031-16980-9_7 10.1016/j.radonc.2020.06.049 10.1088/1361‐6560/abf1bb 10.1038/s41551‐024‐01283‐7 10.1002/mp.16704 10.1002/mp.12251 10.1016/j.media.2023.103046 10.1109/ICCV.2017.244 10.1109/CVPR.2018.00068 10.1016/j.rpor.2019.02.001 10.1088/1361-6560/abd953 10.1109/CVPR52688.2022.01042 10.1016/j.compbiomed.2023.107054 10.1002/mp.12748 10.1002/mp.14121 10.1002/mp.15264 10.1109/TMI.2023.3325703 10.1002/mp.13927 10.1088/2057‐1976/ab6e1f 10.1109/CISP-BMEI.2018.8633142 10.48550/arXiv.2010.02502 10.1109/CVPR42600.2020.00821 10.3389/fonc.2024.1440944 10.1109/CVPR.2017.632 |
| ContentType | Journal Article |
| Copyright | 2025 American Association of Physicists in Medicine. |
| Copyright_xml | – notice: 2025 American Association of Physicists in Medicine. |
| DBID | AAYXX CITATION CGR CUY CVF ECM EIF NPM 7X8 |
| DOI | 10.1002/mp.17912 |
| DatabaseName | CrossRef Medline MEDLINE MEDLINE (Ovid) MEDLINE MEDLINE PubMed MEDLINE - Academic |
| DatabaseTitle | CrossRef MEDLINE Medline Complete MEDLINE with Full Text PubMed MEDLINE (Ovid) MEDLINE - Academic |
| DatabaseTitleList | MEDLINE MEDLINE - Academic |
| Database_xml | – sequence: 1 dbid: NPM name: PubMed url: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed sourceTypes: Index Database – sequence: 2 dbid: 7X8 name: MEDLINE - Academic url: https://search.proquest.com/medline sourceTypes: Aggregation Database |
| DeliveryMethod | fulltext_linktorsrc |
| Discipline | Medicine Physics |
| EISSN | 2473-4209 |
| EndPage | n/a |
| ExternalDocumentID | 40439703 10_1002_mp_17912 MP17912 |
| Genre | researchArticle Journal Article |
| GrantInformation_xml | – fundername: Natural Science Foundation of Hubei Province of China funderid: 2025AFB842 – fundername: Natural Science Foundation of Hubei Province of China grantid: 2025AFB842 |
| GroupedDBID | --- --Z -DZ .GJ 0R~ 1OB 1OC 29M 2WC 33P 36B 3O- 4.4 53G 5GY 5RE 5VS AAHQN AAIPD AAMMB AAMNL AANLZ AAQQT AASGY AAXRX AAYCA AAZKR ABCUV ABDPE ABEFU ABFTF ABJNI ABLJU ABQWH ABXGK ACAHQ ACBEA ACCZN ACGFO ACGFS ACGOF ACPOU ACXBN ACXQS ADBBV ADBTR ADKYN ADMLS ADOZA ADXAS ADZMN AEFGJ AEGXH AEIGN AENEX AEUYR AEYWJ AFBPY AFFPM AFWVQ AGHNM AGXDD AGYGG AHBTC AIACR AIAGR AIDQK AIDYY AITYG AIURR ALMA_UNASSIGNED_HOLDINGS ALUQN ALVPJ AMYDB ASPBG BFHJK C45 CS3 DCZOG DRFUL DRMAN DRSTM DU5 EBD EBS EJD EMB EMOBN F5P HDBZQ HGLYW I-F KBYEO LATKE LEEKS LOXES LUTES LYRES MEWTI O9- OVD P2P P2W PALCI PHY RJQFR RNS ROL SAMSI SUPJJ SV3 TEORI TN5 TWZ USG WOHZO WXSBR XJT ZGI ZVN ZXP ZY4 ZZTAW AAYXX ABUFD AIQQE CITATION LH4 CGR CUY CVF ECM EIF NPM 7X8 |
| ID | FETCH-LOGICAL-c2842-65c5460c3cfc84bb4110aeaf323eb874b4d1ff8be70337549a0c81785cf39a103 |
| IEDL.DBID | DRFUL |
| ISICitedReferencesCount | 1 |
| ISICitedReferencesURI | http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001498563100001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D |
| ISSN | 0094-2405 2473-4209 |
| IngestDate | Fri Sep 05 16:02:48 EDT 2025 Wed Jul 16 06:54:32 EDT 2025 Sat Nov 29 07:29:51 EST 2025 Tue Jul 22 09:32:03 EDT 2025 |
| IsPeerReviewed | true |
| IsScholarly | true |
| Issue | 7 |
| Keywords | deep learning multimodality image translation latent space mapping bidirectional transformer adaptive radiotherapy |
| Language | English |
| License | 2025 American Association of Physicists in Medicine. |
| LinkModel | DirectLink |
| MergedId | FETCHMERGED-LOGICAL-c2842-65c5460c3cfc84bb4110aeaf323eb874b4d1ff8be70337549a0c81785cf39a103 |
| Notes | Zhiwen Liang and Mengjie Cheng contributed equally to this work. ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 23 |
| PMID | 40439703 |
| PQID | 3213612289 |
| PQPubID | 23479 |
| PageCount | 12 |
| ParticipantIDs | proquest_miscellaneous_3213612289 pubmed_primary_40439703 crossref_primary_10_1002_mp_17912 wiley_primary_10_1002_mp_17912_MP17912 |
| PublicationCentury | 2000 |
| PublicationDate | July 2025 |
| PublicationDateYYYYMMDD | 2025-07-01 |
| PublicationDate_xml | – month: 07 year: 2025 text: July 2025 |
| PublicationDecade | 2020 |
| PublicationPlace | United States |
| PublicationPlace_xml | – name: United States |
| PublicationTitle | Medical physics (Lancaster) |
| PublicationTitleAlternate | Med Phys |
| PublicationYear | 2025 |
| References | 2023; 51 2020; 6 2021; 48 2021; 66 2023 2022 2023; 162 2025; 9 2020 2017; 44 2019; 24 2024; 92 2020; 150 2018 2017 2020; 47 2024; 43 2020; 33 2024 2021; 1 2018; 45 2024; 14 e_1_2_8_28_1 e_1_2_8_29_1 e_1_2_8_24_1 e_1_2_8_25_1 e_1_2_8_26_1 e_1_2_8_27_1 Ho J (e_1_2_8_12_1) 2020 e_1_2_8_3_1 e_1_2_8_2_1 e_1_2_8_5_1 e_1_2_8_4_1 e_1_2_8_7_1 e_1_2_8_6_1 e_1_2_8_9_1 Karras T (e_1_2_8_11_1) 2020 e_1_2_8_8_1 e_1_2_8_20_1 e_1_2_8_21_1 e_1_2_8_22_1 e_1_2_8_23_1 e_1_2_8_17_1 e_1_2_8_18_1 e_1_2_8_39_1 e_1_2_8_19_1 e_1_2_8_13_1 e_1_2_8_36_1 e_1_2_8_14_1 e_1_2_8_35_1 e_1_2_8_15_1 e_1_2_8_38_1 e_1_2_8_16_1 e_1_2_8_37_1 e_1_2_8_32_1 e_1_2_8_10_1 e_1_2_8_31_1 e_1_2_8_34_1 e_1_2_8_33_1 e_1_2_8_30_1 |
| References_xml | – volume: 24 start-page: 245 issue: 2 year: 2019 end-page: 250 article-title: Validation of dose distribution computation on sCT images generated from MRI scans by Philips MRCAT publication-title: Rep Pract Oncol Radiother – volume: 14 year: 2024 article-title: A joint learning framework for multisite CBCT‐to‐CT translation using a hybrid CNN‐transformer synthesizer and a registration network publication-title: Front Oncol – volume: 9 start-page: 521 issue: 4 year: 2025 end-page: 538 article-title: A foundation model for enhancing magnetic resonance images and downstream segmentation, registration and diagnostic tasks publication-title: Nat Biomed Eng – start-page: 8185 year: 2020 end-page: 8194 article-title: StarGAN v2: diverse image synthesis for multiple domains – volume: 1 start-page: 5 issue: 1 year: 2021 article-title: When medical images meet generative adversarial network: recent development and research opportunities publication-title: Discov Artif Intell – volume: 6 issue: 1 year: 2020 article-title: Generation of abdominal synthetic CTs from 0.35T MR images using generative adversarial networks for MR‐only liver radiotherapy publication-title: Biomed Phys Eng Express – year: 2017 article-title: Image‐to‐image translation with conditional adversarial networks – year: 2024 – year: 2020 article-title: Denoising diffusion implicit models publication-title: ArXiv E‐Prints – start-page: 586 year: 2018 end-page: 595 article-title: The unreasonable effectiveness of deep features as a perceptual metric – start-page: 66 year: 2022 end-page: 78 article-title: Morphology‐Preserving Autoregressive 3D Generative Modelling of the Brain – start-page: 2242 year: 2017 end-page: 2251 article-title: Unpaired image‐to‐image translation using cycle‐consistent adversarial networks – volume: 150 start-page: 217 year: 2020 end-page: 224 article-title: Magnetic resonance‐based synthetic computed tomography images generated using generative adversarial networks for nasopharyngeal carcinoma radiotherapy treatment planning publication-title: Radiother Oncol – volume: 43 start-page: 980 issue: 3 year: 2024 end-page: 993 article-title: Zero‐shot medical image translation via frequency‐guided diffusion models publication-title: IEEE Trans Med Imaging – start-page: 1 year: 2018 end-page: 5 article-title: A novel method of synthetic CT generation from MR images based on convolutional neural networks – start-page: 10674 year: 2022 end-page: 10685 article-title: High‐Resolution Image Synthesis with Latent Diffusion Models – volume: 45 start-page: 1295 issue: 3 year: 2018 end-page: 1300 article-title: MR and CT data with multiobserver delineations of organs in the pelvic area—Part of the Gold Atlas project publication-title: Med Phys – volume: 48 start-page: 7063 issue: 11 year: 2021 end-page: 7073 article-title: Synthetic CT‐aided multiorgan segmentation for CBCT‐guided adaptive pancreatic radiotherapy publication-title: Med Phys – volume: 92 year: 2024 article-title: Deep learning based synthesis of MRI, CT and PET: review and analysis publication-title: Med Image Anal – volume: 47 start-page: 626 issue: 2 year: 2020 end-page: 642 article-title: Patch‐based generative adversarial neural network models for head and neck MR‐only planning publication-title: Med Phys – year: 2022 article-title: DPM‐solver: A Fast ODE solver for diffusion probabilistic model sampling in around 10 steps – year: 2022 – volume: 47 start-page: 2472 issue: 6 year: 2020 end-page: 2483 article-title: CBCT‐based synthetic CT generation using deep‐attention cycleGAN for pancreatic adaptive radiotherapy publication-title: Med Phys – year: 2020 – year: 2023 – volume: 162 year: 2023 article-title: Multimodality MRI synchronous construction based deep learning framework for MRI‐guided radiotherapy synthetic CT generation publication-title: Comput Biol Med – volume: 33 start-page: 6840 year: 2020 end-page: 6851 – volume: 51 start-page: 1847 issue: 3 year: 2023 end-page: 1859 article-title: CBCT‐Based synthetic CT image generation using conditional denoising diffusion probabilistic model publication-title: Med Phys – volume: 66 issue: 9 year: 2021 article-title: A feature invariant generative adversarial network for head and neck MRI/CT image synthesis publication-title: Phys Med Biol – volume: 44 start-page: 2556 issue: 6 year: 2017 end-page: 2568 article-title: Development of the open‐source dose calculation and optimization toolkit matRad publication-title: Med Phys – start-page: 1 year: 2020 end-page: 11 – ident: e_1_2_8_5_1 doi: 10.1007/s44163‐021‐00006‐0 – start-page: 6840 volume-title: Advances in Neural Information Processing Systems year: 2020 ident: e_1_2_8_12_1 – ident: e_1_2_8_37_1 doi: 10.1007/978-3-031-16980-9_7 – ident: e_1_2_8_14_1 – ident: e_1_2_8_8_1 doi: 10.1016/j.radonc.2020.06.049 – ident: e_1_2_8_17_1 – ident: e_1_2_8_4_1 doi: 10.1088/1361‐6560/abf1bb – ident: e_1_2_8_22_1 – ident: e_1_2_8_39_1 – ident: e_1_2_8_25_1 – ident: e_1_2_8_35_1 doi: 10.1038/s41551‐024‐01283‐7 – ident: e_1_2_8_3_1 doi: 10.1002/mp.16704 – ident: e_1_2_8_24_1 doi: 10.1002/mp.12251 – ident: e_1_2_8_2_1 doi: 10.1016/j.media.2023.103046 – ident: e_1_2_8_6_1 doi: 10.1109/ICCV.2017.244 – ident: e_1_2_8_19_1 doi: 10.1109/CVPR.2018.00068 – ident: e_1_2_8_21_1 – ident: e_1_2_8_29_1 doi: 10.1016/j.rpor.2019.02.001 – start-page: 1 volume-title: Proceedings of the 34th International Conference on Neural Information Processing Systems (NeurIPS 2020) year: 2020 ident: e_1_2_8_11_1 – ident: e_1_2_8_33_1 doi: 10.1088/1361-6560/abd953 – ident: e_1_2_8_16_1 doi: 10.1109/CVPR52688.2022.01042 – ident: e_1_2_8_28_1 doi: 10.1016/j.compbiomed.2023.107054 – ident: e_1_2_8_23_1 doi: 10.1002/mp.12748 – ident: e_1_2_8_27_1 – ident: e_1_2_8_31_1 doi: 10.1002/mp.14121 – ident: e_1_2_8_34_1 doi: 10.1002/mp.15264 – ident: e_1_2_8_38_1 – ident: e_1_2_8_30_1 – ident: e_1_2_8_13_1 doi: 10.1109/TMI.2023.3325703 – ident: e_1_2_8_15_1 – ident: e_1_2_8_9_1 doi: 10.1002/mp.13927 – ident: e_1_2_8_7_1 doi: 10.1088/2057‐1976/ab6e1f – ident: e_1_2_8_10_1 doi: 10.1109/CISP-BMEI.2018.8633142 – ident: e_1_2_8_18_1 doi: 10.48550/arXiv.2010.02502 – ident: e_1_2_8_32_1 doi: 10.1109/CVPR42600.2020.00821 – ident: e_1_2_8_36_1 doi: 10.3389/fonc.2024.1440944 – ident: e_1_2_8_20_1 doi: 10.1109/CVPR.2017.632 – ident: e_1_2_8_26_1 |
| SSID | ssj0006350 |
| Score | 2.4769492 |
| Snippet | Background
Medical image translation has become an essential tool in modern radiotherapy, providing complementary information for target delineation and dose... Medical image translation has become an essential tool in modern radiotherapy, providing complementary information for target delineation and dose calculation.... |
| SourceID | proquest pubmed crossref wiley |
| SourceType | Aggregation Database Index Database Publisher |
| StartPage | e17912 |
| SubjectTerms | adaptive radiotherapy Autoencoder bidirectional transformer Cone-Beam Computed Tomography deep learning Head and Neck Neoplasms - diagnostic imaging Humans Image Processing, Computer-Assisted - methods latent space mapping Magnetic Resonance Imaging Multimodal Imaging - methods multimodality image translation |
| Title | Multimodal medical image‐to‐image translation via variational autoencoder latent space mapping |
| URI | https://onlinelibrary.wiley.com/doi/abs/10.1002%2Fmp.17912 https://www.ncbi.nlm.nih.gov/pubmed/40439703 https://www.proquest.com/docview/3213612289 |
| Volume | 52 |
| WOSCitedRecordID | wos001498563100001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D |
| hasFullText | 1 |
| inHoldings | 1 |
| isFullTextHit | |
| isPrint | |
| journalDatabaseRights | – providerCode: PRVWIB databaseName: Wiley Online Library Full Collection 2020 customDbUrl: eissn: 2473-4209 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0006350 issn: 0094-2405 databaseCode: DRFUL dateStart: 19970101 isFulltext: true titleUrlDefault: https://onlinelibrary.wiley.com providerName: Wiley-Blackwell |
| link | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwpV1LT-MwEB6xZRdxAZblUV7ySqu9BepHGvuIgIpDi9BqWfUW2Y6DekhTtWnP_AR-I7-EsZN2hVZIK3FJFGliWZ7xzNgz8w3Aj0S7JNPKRTzWMhJdoSNjqc8yd1Qb2s1dgGP400_u7uRwqO6brEpfC1PjQ6wu3PzOCPrab3BtZhd_QUOLybmH1kT1u85QbEUL1q9_9R76Kz2MprQuQFHCxxDiJfRsh10s_31rjP7xMN86rMHi9LY_Mtcd2Gr8THJZC8ZXWHPjXdgYNJH0XfgSUj_t7BuYUINblBmSF3XYhowKVDMvT89ViY_wQSpv1OrEObIYabLAQ3ZzkUj0vCo9IGbmpgRJ0I4RVFTWkUJ7-IfHPXjo3fy-uo2azguRRXPFom5sY9HtWG5zK4UxAp0E7XTOGXdGJsKIjOa5NA4X3vfQVbpjJU1kbHOuNO3wfWiNy7E7BJJT9AkY5yLWVMTWSeoDtYZmVDmmMtWG70sWpJMaYCOtoZRZWkzSsGxIs-RNitLvQxp67Mr5LOWMcvTR8NTYhoOaaatRPG6Qwgm24WfgzbvDp4P78D76X8Jj2GS-DXDI2j2BVjWdu1P4bBfVaDY9g0_JUJ410vgKf03ibQ |
| linkProvider | Wiley-Blackwell |
| linkToHtml | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwpV1fa9swED9C2m19Wdus27K1qwZjb24tS44t9jTaho4mIYxk9M1Islzy4DgkTp77EfYZ90l2ku2MMAaFvdgYzkLoTnen-_MTwKdImiiVwngslLHHe1x6SlNbZW6oVLSXGQfH8GMQjUbx_b0Yt-BL0wtT4UNsA252Zzh9bTe4DUhf_kENzRcXFlsT9e8eRykK27B3_b0_HWwVMdrSqgNFcJtECBvsWT-4bP7dtUZ_uZi7HqszOf3D_5rsEbysPU3ytRKNY2iZeQeeD-tcegeeueJPvXoFynXh5kWK5HmVuCGzHBXNr8efZYEP90FKa9aq0jmymUmywWN2HUokcl0WFhIzNUuCJGjJCKoqbUguLQDEwwlM-zeTq1uvvnvB02iwAq8X6pD3fM10pmOuFEc3QRqZsYAZFUdc8ZRmWawMagx7i66Qvo5pFIc6Y0JSn72G9ryYm7dAMopeQcAYDyXloTYxtalaRVMqTCBS0YWPDQ-SRQWxkVRgykGSLxK3bEjTMCdB-bdJDTk3xXqVsIAy9NLw3NiFNxXXtqNY5CCBE-zCZ8ecfw6fDMfu_e6phOfw4nYyHCSDb6O793AQ2EuBXQ3vKbTL5dqcwb7elLPV8kMtlL8Bt4vldQ |
| linkToPdf | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwpV1Li9swEB5C0g29bNts201fq0Lpzd3Ikh-ip9I0tGwSwrIpuRlJlksOjkPi5Lw_YX9jf0lHsp0llMJCLzaGsRCapzQznwA-RNJEqRTGY4GMPR5y6SlNbZW5oVLRMDMOjuHnOJpO48VCzFrwuemFqfAhDgduVjOcvbYKbtZpdnmPGpqvP1lsTbS_HR6IELWyM7wezccHQ4y-tOpAEdwmEYIGe3bgXzb_Hnujv0LM44jVuZzRk_-a7FM4rSNN8qUSjWfQMqsedCd1Lr0HJ674U2_PQLku3LxIkTyvEjdkmaOh-X17Vxb4cB-ktG6tKp0j-6Uke9xm10eJRO7KwkJipmZDkAQ9GUFTpQ3JpQWA-PUc5qNvN1-_e_XdC55Gh-V7YaADHg4005mOuVIcwwRpZMZ8ZlQcccVTmmWxMmgx7C26Qg50TKM40BkTkg7YC2ivipU5B5JRjAp8xnggKQ-0ialN1SqaUmF8kYo-vG94kKwriI2kAlP2k3yduGVDmoY5Ccq_TWrIlSl224T5lGGUhvvGPrysuHYYxSIHCZxgHz465vxz-GQyc-9XDyW8gO5sOErGP6ZXr-Gxb-8EdiW8b6BdbnbmLTzS-3K53byrZfIParvk8A |
| openUrl | ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Multimodal+medical+image-to-image+translation+via+variational+autoencoder+latent+space+mapping&rft.jtitle=Medical+physics+%28Lancaster%29&rft.au=Liang%2C+Zhiwen&rft.au=Cheng%2C+Mengjie&rft.au=Ma%2C+Jinhui&rft.au=Hu%2C+Ying&rft.date=2025-07-01&rft.eissn=2473-4209&rft.volume=52&rft.issue=7&rft.spage=e17912&rft_id=info:doi/10.1002%2Fmp.17912&rft_id=info%3Apmid%2F40439703&rft.externalDocID=40439703 |
| thumbnail_l | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=0094-2405&client=summon |
| thumbnail_m | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=0094-2405&client=summon |
| thumbnail_s | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=0094-2405&client=summon |