{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,20]],"date-time":"2026-05-20T19:16:05Z","timestamp":1779304565219,"version":"3.51.4"},"reference-count":42,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,3,12]],"date-time":"2026-03-12T00:00:00Z","timestamp":1773273600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Neuroinform."],"abstract":"<jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>Secondary quantitative analysis of brain magnetic resonance imaging (MRI) can provide valuable information for many neurological diseases, including multiple sclerosis (MS), but it demands complete datasets that are often unavailable clinically. We investigated how image synthesis via deep learning using cycle-consistent generative adversarial networks (CycleGANs) compared with Pix2Pix as a related method, based on T1-weighted and T2-weighted brain MRI in MS, following verification on two streamlined datasets. The synthesized images were also evaluated against the source data.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>The streamlined datasets involved 1,113 healthy participants from the Human Connectome Project (HCP) and 318 participants from the Parkinson\u2019s Progression Markers Initiative (PPMI). The MS cohort in this study included 105 participants scanned with different protocols. Image synthesis was bidirectional between T1- and T2-weighted MRI using CycleGAN with and without spectral normalization, as well as Pix2Pix. Utility testing focused on T1-weighted MRI that was most often unavailable in MS, and that involved lesion detection, brain volumetry, and lesion texture analysis.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>\n                      All CycleGAN models performed competitively, while Pix2Pix performed better, mostly with streamlined datasets (\n                      <jats:italic>p<\/jats:italic>\n                      \u202f&amp;lt;\u202f0.001). The average peak signal-to-noise ratio ranged from 24.860\u201328.570 versus 28.520\u201331.100, and the structural similarity index ranged from 0.838\u20130.901 versus 0.924\u20130.943. With spectral normalization, CycleGAN improved in PPMI but not in HCP and generally not in MS (\n                      <jats:italic>p<\/jats:italic>\n                      \u202f&amp;lt;\u202f0.001). Furthermore, the synthesized images showed high similarity to the source data in utility tests, although Pix2Pix T1 images appeared more heterogeneous in lesion texture than source T1 images.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>CycleGAN without spectral normalization appeared feasible for synthesizing common clinical brain MRI, including T1-weighted images usable for subsequent quantitative analysis in MS.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/fninf.2026.1762794","type":"journal-article","created":{"date-parts":[[2026,3,12]],"date-time":"2026-03-12T06:49:42Z","timestamp":1773298182000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":1,"title":["CycleGAN models show consistent brain MRI synthesis across datasets supporting downstream tissue characterization in multiple sclerosis"],"prefix":"10.3389","volume":"20","author":[{"given":"Shayan","family":"Shahrokhi","sequence":"first","affiliation":[{"name":"Neuroscience Graduate Program, Faculty of Graduate Studies, University of Calgary","place":["Calgary, AB, Canada"]},{"name":"Hotchkiss Brain Institute, Cumming School of Medicine, University of Calgary","place":["Calgary, AB, Canada"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Olayinka","family":"Oladosu","sequence":"additional","affiliation":[{"name":"Hotchkiss Brain Institute, Cumming School of Medicine, University of Calgary","place":["Calgary, AB, Canada"]},{"name":"Department of Radiology, University of Calgary","place":["Calgary, AB, Canada"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Rehman","family":"Tariq","sequence":"additional","affiliation":[{"name":"Biomedical Engineering Graduate Program, Faculty of Graduate Studies, University of Calgary","place":["Calgary, AB, Canada"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Yunyan","family":"Zhang","sequence":"additional","affiliation":[{"name":"Hotchkiss Brain Institute, Cumming School of Medicine, University of Calgary","place":["Calgary, AB, Canada"]},{"name":"Department of Radiology, University of Calgary","place":["Calgary, AB, Canada"]},{"name":"Department of Clinical Neurosciences, University of Calgary","place":["Calgary, AB, Canada"]}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"1965","published-online":{"date-parts":[[2026,3,12]]},"reference":[{"key":"ref1","doi-asserted-by":"publisher","first-page":"265","DOI":"10.48550\/arXiv.1605.08695","article-title":"TensorFlow: a system for large-scale machine learning","volume":"16","author":"Abadi","year":"2016","journal-title":"arXiv"},{"key":"ref2","first-page":"214","author":"Arjovsky","year":"2017"},{"key":"ref3","volume-title":"Simulation and Synthesis in Medical Imaging","author":"Basaran","year":"2022"},{"key":"ref4","first-page":"728","author":"Bui","year":"2020"},{"key":"ref5","doi-asserted-by":"publisher","first-page":"e210097","DOI":"10.1148\/ryai.2021210097","article-title":"Toward generalizability in the deployment of artificial intelligence in radiology: role of computation stress testing to overcome underspecification","volume":"3","author":"Eche","year":"2021","journal-title":"Radiol. Artif. Intell."},{"key":"ref6","doi-asserted-by":"publisher","first-page":"889808","DOI":"10.3389\/fnins.2022.889808","article-title":"Uncertainty-aware and lesion-specific image synthesis in multiple sclerosis magnetic resonance imaging: a multicentric validation study","volume":"16","author":"Finck","year":"2022","journal-title":"Front. Neurosci."},{"key":"ref7","doi-asserted-by":"publisher","first-page":"774","DOI":"10.1016\/j.neuroimage.2012.01.021","article-title":"Freesurfer","volume":"62","author":"Fischl","year":"2012","journal-title":"NeuroImage"},{"key":"ref8","doi-asserted-by":"publisher","first-page":"808","DOI":"10.1093\/brain\/awm329","article-title":"Disability and T2 MRI lesions: a 20-year follow-up of patients with relapse onset of multiple sclerosis","volume":"131","author":"Fisniku","year":"2008","journal-title":"Brain"},{"key":"ref9","doi-asserted-by":"publisher","first-page":"139","DOI":"10.1145\/3422622","article-title":"Generative adversarial networks","volume":"63","author":"Goodfellow","year":"2020","journal-title":"Commun. ACM"},{"key":"ref10","doi-asserted-by":"publisher","first-page":"109671","DOI":"10.1016\/j.jneumeth.2022.109671","article-title":"Texture analysis in brain T2 and diffusion MRI differentiates histology-verified grey and white matter pathology types in multiple sclerosis","volume":"379","author":"Hosseinpour","year":"2022","journal-title":"J. Neurosci. Methods"},{"key":"ref11","doi-asserted-by":"publisher","first-page":"161107004v3","DOI":"10.1109\/CVPR.2017.632","article-title":"Image-to-image translation with conditional adversarial networks","author":"Isola","year":"2016","journal-title":"arXiv"},{"key":"ref12","first-page":"1125","author":"Isola","year":"2017"},{"key":"ref13","doi-asserted-by":"publisher","first-page":"782","DOI":"10.1016\/j.neuroimage.2011.09.015","article-title":"FSL","volume":"62","author":"Jenkinson","year":"2012","journal-title":"NeuroImage"},{"key":"ref14","author":"Juli\u00e1n Alberto","year":"2020"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1412.6980","article-title":"Adam: a method for stochastic optimization","author":"Kingma","year":"2014","journal-title":"arXiv"},{"key":"ref16","doi-asserted-by":"publisher","first-page":"e232471","DOI":"10.1148\/radiol.232471","article-title":"Generating synthetic data for medical imaging","volume":"312","author":"Koetzier","year":"2024","journal-title":"Radiology"},{"key":"ref17","doi-asserted-by":"publisher","first-page":"104297","DOI":"10.1016\/j.compbiomed.2021.104297","article-title":"MPRAGE to MP2RAGE UNI translation via generative adversarial network improves the automatic tissue and lesion segmentation in multiple sclerosis patients","volume":"132","author":"La Rosa","year":"2021","journal-title":"Comput. Biol. Med."},{"key":"ref18","doi-asserted-by":"publisher","first-page":"4528","DOI":"10.3390\/app10134528","article-title":"Improvement of learning stability of generative adversarial network using variational learning","volume":"10","author":"Lee","year":"2020","journal-title":"Appl. Sci."},{"key":"ref19","first-page":"795","volume-title":"Lecture Notes in Computer Science","author":"Li","year":"2019"},{"key":"ref20","doi-asserted-by":"publisher","first-page":"9625","DOI":"10.48550\/arXiv.2009.02773","article-title":"Why spectral normalization stabilizes GANs: analysis and improvements","volume":"34","author":"Lin","year":"2021","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"ref21","doi-asserted-by":"publisher","first-page":"629","DOI":"10.1016\/j.pneurobio.2011.09.005","article-title":"The Parkinson progression marker initiative (PPMI)","volume":"95","author":"Marek","year":"2011","journal-title":"Prog. Neurobiol."},{"key":"ref22","doi-asserted-by":"publisher","first-page":"765","DOI":"10.1001\/jama.2020.26858","article-title":"Diagnosis and treatment of multiple sclerosis","volume":"325","author":"McGinley","year":"2021","journal-title":"JAMA"},{"key":"ref23","doi-asserted-by":"publisher","first-page":"101641","DOI":"10.1016\/j.bspc.2019.101641","article-title":"An unpaired GAN for brain tumor segmentation","volume":"55","author":"Nema","year":"2020","journal-title":"Biomed. Signal Process. Control"},{"key":"ref24","doi-asserted-by":"publisher","first-page":"e1561","DOI":"10.1212\/WNL.0000000000011494","article-title":"Association of gray matter atrophy patterns with clinical phenotype and progression in multiple sclerosis","volume":"96","author":"Rocca","year":"2021","journal-title":"Neurology"},{"key":"ref25","volume-title":"Medical Image Computing and Computer Assisted Intervention\u2014MICCAI 2024 Workshops","author":"Rokuss","year":"2025"},{"key":"ref26","doi-asserted-by":"publisher","first-page":"S208","DOI":"10.1016\/j.neuroimage.2004.07.051","article-title":"Advances in functional and structural mr image analysis and implementation as fsl","volume":"23","author":"Smith","year":"2004","journal-title":"NeuroImage"},{"key":"ref27","doi-asserted-by":"publisher","first-page":"120494","DOI":"10.1016\/j.neuroimage.2023.120494","article-title":"Accuracy of TrUE-net in comparison to established white matter hyperintensity segmentation methods: an independent validation study","volume":"285","author":"Strain","year":"2024","journal-title":"NeuroImage"},{"key":"ref28","doi-asserted-by":"publisher","first-page":"1622","DOI":"10.1016\/S0140-6736(18)30481-1","article-title":"Multiple sclerosis","volume":"391","author":"Thompson","year":"2018","journal-title":"Lancet"},{"key":"ref29","doi-asserted-by":"publisher","first-page":"1310","DOI":"10.1109\/TMI.2010.2046908","article-title":"N4ITK: improved N3 Bias correction","volume":"29","author":"Tustison","year":"2010","journal-title":"IEEE Trans. Med. Imaging"},{"key":"ref30","doi-asserted-by":"publisher","first-page":"2222","DOI":"10.1016\/j.neuroimage.2012.01.032","article-title":"The future of the human connectome","volume":"62","author":"Van Essen","year":"2012","journal-title":"NeuroImage"},{"key":"ref31","doi-asserted-by":"publisher","first-page":"577","DOI":"10.1016\/j.nrleng.2020.10.013","article-title":"A systematic review of the application of machine-learning algorithms in multiple sclerosis","volume":"38","author":"V\u00e1zquez-Marrufo","year":"2023","journal-title":"Neurologia"},{"key":"ref32","doi-asserted-by":"publisher","first-page":"11","DOI":"10.1002\/acm2.13121","article-title":"A review on medical imaging synthesis using deep learning and its clinical applications","volume":"22","author":"Wang","year":"2021","journal-title":"J. Appl. Clin. Med. Phys."},{"key":"ref33","doi-asserted-by":"publisher","first-page":"653","DOI":"10.1016\/S1474-4422(21)00095-8","article-title":"2021 MAGNIMS-CMSC-NAIMS consensus recommendations on the use of MRI in patients with multiple sclerosis","volume":"20","author":"Wattjes","year":"2021","journal-title":"Lancet Neurol."},{"key":"ref34","doi-asserted-by":"publisher","first-page":"180607777","DOI":"10.48550\/arXiv.1806.07777","article-title":"Generative adversarial networks for image-to-image translation on multi-contrast mr images\u2014a comparison of CycleGAN and unit","author":"Welander","year":"2018","journal-title":"arXiv"},{"key":"ref35","doi-asserted-by":"publisher","first-page":"82","DOI":"10.1016\/j.neunet.2020.05.001","article-title":"BPGAN: bidirectional CT-to-MRI prediction using multi-generative multi-adversarial nets with spectral normalization and localization","volume":"128","author":"Xu","year":"2020","journal-title":"Neural Netw."},{"key":"ref36","doi-asserted-by":"publisher","first-page":"103994","DOI":"10.1016\/j.bspc.2022.103994","article-title":"Bi-MGAN: bidirectional T1-to-T2 MRI images prediction using multi-generative multi-adversarial nets","volume":"78","author":"Xu","year":"2022","journal-title":"Biomed. Signal Process. Control"},{"key":"ref37","doi-asserted-by":"publisher","first-page":"3753","DOI":"10.1038\/s41598-020-60520-6","article-title":"MRI cross-modality image-to-image translation","volume":"10","author":"Yang","year":"2020","journal-title":"Sci. Rep."},{"key":"ref38","doi-asserted-by":"publisher","first-page":"109098","DOI":"10.1016\/j.jneumeth.2021.109098","article-title":"Grad-CAM helps interpret the deep learning models trained to classify multiple sclerosis types using clinical brain magnetic resonance imaging","volume":"353","author":"Zhang","year":"2021","journal-title":"J. Neurosci. Methods"},{"key":"ref39","doi-asserted-by":"publisher","first-page":"816","DOI":"10.3390\/diagnostics12040816","article-title":"Multi-contrast MRI image synthesis using switchable cycle-consistent generative adversarial networks","volume":"12","author":"Zhang","year":"2022","journal-title":"Diagnostics (Basel)"},{"key":"ref40","volume-title":"2018 Digital Image Computing: Techniques and Applications (DICTA)","author":"Zhang","year":"2018"},{"key":"ref41","author":"Zhao","year":"2021"},{"key":"ref42","author":"Zhu","year":"2017"}],"container-title":["Frontiers in Neuroinformatics"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fninf.2026.1762794\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,12]],"date-time":"2026-03-12T06:49:44Z","timestamp":1773298184000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fninf.2026.1762794\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,3,12]]},"references-count":42,"alternative-id":["10.3389\/fninf.2026.1762794"],"URL":"https:\/\/doi.org\/10.3389\/fninf.2026.1762794","relation":{},"ISSN":["1662-5196"],"issn-type":[{"value":"1662-5196","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,3,12]]},"article-number":"1762794"}}