{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,15]],"date-time":"2026-09-15T17:07:00Z","timestamp":1789492020943,"version":"build-2803163510"},"reference-count":53,"publisher":"Wiley","issue":"5","license":[{"start":{"date-parts":[[2026,5,21]],"date-time":"2026-05-21T00:00:00Z","timestamp":1779321600000},"content-version":"vor","delay-in-days":20,"URL":"http:\/\/onlinelibrary.wiley.com\/termsAndConditions#vor"},{"start":{"date-parts":[[2026,5,1]],"date-time":"2026-05-01T00:00:00Z","timestamp":1777593600000},"content-version":"tdm","delay-in-days":0,"URL":"http:\/\/doi.wiley.com\/10.1002\/tdm_license_1.1"}],"funder":[{"DOI":"10.13039\/100000002","name":"National Institutes of Health","doi-asserted-by":"publisher","award":["R01EB032680"],"award-info":[{"award-number":["R01EB032680"]}],"id":[{"id":"10.13039\/100000002","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100000002","name":"National Institutes of Health","doi-asserted-by":"publisher","award":["R01CA272991"],"award-info":[{"award-number":["R01CA272991"]}],"id":[{"id":"10.13039\/100000002","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100000002","name":"National Institutes of Health","doi-asserted-by":"publisher","award":["R21EB033994"],"award-info":[{"award-number":["R21EB033994"]}],"id":[{"id":"10.13039\/100000002","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["aapm.onlinelibrary.wiley.com"],"crossmark-restriction":true},"short-container-title":["Medical Physics"],"published-print":{"date-parts":[[2026,5]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>Ultra\u2010high\u2010field 7 Tesla (7T) magnetic resonance imaging (MRI) provides improved resolution and signal\u2010to\u2010noise ratio (SNR) over standard clinical field strengths (1.5T, 3T). However, 7T scanners are costly, scarce, and introduce additional challenges such as susceptibility artifacts.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Purpose<\/jats:title>\n                    <jats:p>We propose an efficient transformer\u2010based model (7T\u2010Restormer) to synthesize 7T\u2010like MRI (quantitative T1 MP2RAGE maps) from routine 1.5T or 3T T1\u2010weighted (T1W) images.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>\n                      The proposed method leverages an efficient restoration transformer backbone and spatial attention layers to capture long\u2010range dependencies and generate high\u2010quality 7T\u2010like images. Our model was validated on an institutional dataset, comprised of 35 1.5T and 108 3T T1w MRI paired with corresponding 7T T1 maps of patients with confirmed multiple sclerosis (MS). A total of 141 patient cases (32\u00a0128 slices) were randomly divided into 105 (25 1.5T and 80 3T) training cases (19\u00a0204 slices), 19 (5 1.5T and 14 3T) validation cases (3476 slices), and 17 (5 1.5T and 12 3T) test cases (3145 slices). The synthetic 7T T1 maps were evaluated by comparing their similarity to the ground truth volumes using peak signal\u2010to\u2010noise ratio (PSNR), structural similarity index measure (SSIM), and normalized mean squared error (NMSE), and were compared against ResViT and ResShift. In addition, a blinded reader study was performed in which three clinicians scored the diagnostic image quality of each method on a 5\u2010point Likert scale (1\u00a0=\u00a0non\u2010acceptable, 5\u00a0=\u00a0excellent). Statistical significance (\n                      <jats:italic>\u03b1<\/jats:italic>\n                      \u00a0=\u00a00.05) was assessed using two\u2010sided paired\n                      <jats:italic>t<\/jats:italic>\n                      \u2010tests with Holm correction for multiple comparisons and Wilcoxon signed\u2010rank tests with Holm correction for reader scores. Effect sizes (Cohen's  for paired data) were also computed to quantify practical significance. We interpreted  and  as small, medium, and large effects.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>\n                      For 1.5T inputs, 7T\u2010Restormer achieved an NMSE of 0.018\u00a0\u00b1\u00a00.003, PSNR of 25.0\u00a0\u00b1\u00a00.6 dB, and SSIM of 0.870\u00a0\u00b1\u00a00.019; for 3T inputs, the corresponding values were 0.019\u00a0\u00b1\u00a00.006, 24.5\u00a0\u00b1\u00a01.2 dB, and 0.874\u00a0\u00b1\u00a00.027. Across both field strengths, 7T\u2010Restormer provided the lowest NMSE and highest PSNR and SSIM among the three methods, reducing NMSE by approximately 5% and 25% relative to ResShift and ResViT at 1.5T and by 14% and 24% at 3T, respectively, and increasing PSNR by 0.5\u20131.5 dB (Holm\u2010adjusted\n                      <jats:italic>p<\/jats:italic>\n                      \u00a0&lt;\u00a00.05, paired Cohen's  =\u00a01.37\u20136.84). Training on mixed 1.5T+3T data improved performance for 1.5T inputs compared with 1.5T\u2010only training (e.g., NMSE 0.018 vs. 0.019;\n                      <jats:italic>p<\/jats:italic>\n                      \u00a0=\u00a00.001, ), without degrading performance for 3T inputs (\n                      <jats:italic>p<\/jats:italic>\n                      \u00a0=\u00a00.630,  for NMSE). In the blinded reader study, 7T\u2010Restormer received higher diagnostic quality scores (3.50\u00a0\u00b1\u00a00.28) than ResShift (3.00\u00a0\u00b1\u00a00.27) and ResViT (2.33\u00a0\u00b1\u00a00.27), corresponding to mean improvements of 0.5\u20131.2 points on the 5\u2010point scale (Holm\u2010adjusted\n                      <jats:italic>p<\/jats:italic>\n                      \u00a0=\u00a00.014 and 0.004; paired Cohen's  and 3.24, respectively).\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>We propose a novel method for predicting quantitative 7T MP2RAGE maps from 1.5T and 3T T1W scans with higher quality than existing state\u2010of\u2010the\u2010art methods. Future development and application of our approach may enhance diagnostic accuracy, treatment planning, and facilitate downstream tasks by making the benefits of 7T MRI more accessible to standard clinical workflows.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1002\/mp.70496","type":"journal-article","created":{"date-parts":[[2026,5,22]],"date-time":"2026-05-22T06:48:26Z","timestamp":1779432506000},"update-policy":"https:\/\/doi.org\/10.1002\/crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Transformer\u2010based synthesis of 7 Tesla\u2010like T1 contrast from routine clinical magnetic resonance imaging"],"prefix":"10.1002","volume":"53","author":[{"given":"Zach","family":"Eidex","sequence":"first","affiliation":[{"name":"Department of Radiation Oncology Emory University  Atlanta Georgia USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Mojtaba","family":"Safari","sequence":"additional","affiliation":[{"name":"Department of Radiation Oncology Emory University  Atlanta Georgia USA"},{"name":"Department of Computer Science and Informatics Emory University  Atlanta Georgia USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Tonghe","family":"Wang","sequence":"additional","affiliation":[{"name":"Department of Medical Physics Memorial Sloan Kettering Cancer Center  New York New York USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"David S.","family":"Yu","sequence":"additional","affiliation":[{"name":"Department of Radiation Oncology Emory University  Atlanta Georgia USA"},{"name":"Winship Cancer Institute Emory University  Atlanta Georgia USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Vanessa","family":"Wildman","sequence":"additional","affiliation":[{"name":"Department of Radiation Oncology Emory University  Atlanta Georgia USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Hui","family":"Mao","sequence":"additional","affiliation":[{"name":"Department of Radiology and Imaging Sciences Emory University  Atlanta Georgia USA"},{"name":"Winship Cancer Institute Emory University  Atlanta Georgia USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Erik","family":"Middlebrooks","sequence":"additional","affiliation":[{"name":"Department of Radiology Mayo Clinic  Jacksonville Florida USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Aparna H.","family":"Kesarwala","sequence":"additional","affiliation":[{"name":"Department of Radiation Oncology Emory University  Atlanta Georgia USA"},{"name":"Winship Cancer Institute Emory University  Atlanta Georgia USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xiaofeng","family":"Yang","sequence":"additional","affiliation":[{"name":"Department of Radiation Oncology Emory University  Atlanta Georgia USA"},{"name":"Department of Computer Science and Informatics Emory University  Atlanta Georgia USA"},{"name":"Winship Cancer Institute Emory University  Atlanta Georgia USA"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"311","published-online":{"date-parts":[[2026,5,21]]},"reference":[{"key":"e_1_2_8_2_1","doi-asserted-by":"publisher","DOI":"10.1259\/bjr.20180492"},{"key":"e_1_2_8_3_1","doi-asserted-by":"publisher","DOI":"10.1002\/acm2.14072"},{"key":"e_1_2_8_4_1","doi-asserted-by":"crossref","unstructured":"SafariM WangS EidexZ et\u00a0al.A Physics\u2010Informed Deep Learning Model for MRI Brain Motion Correction.ArXiv. Preprint posted online February 13 2025.","DOI":"10.1002\/mp.70197"},{"key":"e_1_2_8_5_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.ijcard.2020.10.041"},{"key":"e_1_2_8_6_1","unstructured":"WeissM GeyerS LohmannG TrampelR TurnerR Quantitative T1 mapping at 7 Tesla identifies primary functional areas in the living human brain.2011."},{"key":"e_1_2_8_7_1","doi-asserted-by":"publisher","DOI":"10.1111\/ene.70330"},{"key":"e_1_2_8_8_1","doi-asserted-by":"publisher","DOI":"10.3171\/2023.12.JNS232473"},{"key":"e_1_2_8_9_1","doi-asserted-by":"publisher","DOI":"10.3390\/brainsci15101088"},{"key":"e_1_2_8_10_1","doi-asserted-by":"publisher","DOI":"10.1007\/s10334-025-01237-3"},{"key":"e_1_2_8_11_1","doi-asserted-by":"publisher","DOI":"10.3389\/fneur.2021.789812"},{"key":"e_1_2_8_12_1","doi-asserted-by":"publisher","DOI":"10.1097\/RLI.0000000000000256"},{"key":"e_1_2_8_13_1","doi-asserted-by":"publisher","DOI":"10.1109\/MSP.2019.2950640"},{"key":"e_1_2_8_14_1","doi-asserted-by":"publisher","DOI":"10.1111\/joim.20059"},{"key":"e_1_2_8_15_1","doi-asserted-by":"publisher","DOI":"10.1186\/1532\u2010429X\u201014\u201026"},{"key":"e_1_2_8_16_1","doi-asserted-by":"publisher","DOI":"10.1002\/mp.16847"},{"key":"e_1_2_8_17_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2020.101663"},{"key":"e_1_2_8_18_1","unstructured":"LinK HeckelR Vision Transformers Enable Fast and Robust Accelerated MRI:Proceedings of the 5th International Conference on Medical Imaging with Deep Learning 172:774\u2010795 PMLR;2022"},{"key":"e_1_2_8_19_1","doi-asserted-by":"publisher","DOI":"10.1002\/mp.17600"},{"key":"e_1_2_8_20_1","doi-asserted-by":"publisher","DOI":"10.1002\/acm2.14155"},{"key":"e_1_2_8_21_1","doi-asserted-by":"publisher","DOI":"10.1002\/mp.17079"},{"key":"e_1_2_8_22_1","unstructured":"RombachR BlattmannA LorenzD EsserP OmmerB High\u2010resolution image synthesis with latent diffusion models. arXiv. Preprint posted online April 13 2022. doi:10.48550\/arXiv.2112.10752"},{"key":"e_1_2_8_23_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.bspc.2023.105331"},{"key":"e_1_2_8_24_1","doi-asserted-by":"publisher","DOI":"10.1088\/1361\u20106560\/ad4b91"},{"key":"e_1_2_8_25_1","doi-asserted-by":"publisher","DOI":"10.1002\/mp.17546"},{"key":"e_1_2_8_26_1","doi-asserted-by":"publisher","DOI":"10.1002\/mp.14539"},{"key":"e_1_2_8_27_1","doi-asserted-by":"publisher","DOI":"10.1186\/s12880\u2010023\u201001160\u2010w"},{"key":"e_1_2_8_28_1","doi-asserted-by":"publisher","DOI":"10.1088\/1361\u20106560\/ad4845"},{"key":"e_1_2_8_29_1","doi-asserted-by":"crossref","unstructured":"PanS EidexZ SafariM QiuR YangX Cycle\u2010guided denoising diffusion probability model for 3d cross\u2010modality mri synthesis.Paper presented at: Medical Imaging 2025: Clinical and Biomedical Imaging.SPIE;2025.","DOI":"10.1117\/12.3047506"},{"key":"e_1_2_8_30_1","doi-asserted-by":"publisher","DOI":"10.1088\/1361\u20106560\/ad209c"},{"key":"e_1_2_8_31_1","unstructured":"MojtabaS XiaofengY AliF MRI data consistency guided conditional diffusion probabilistic model for MR imaging acceleration. Paper presented at: Proceedings of SPIE 2024.SPIE;2024."},{"key":"e_1_2_8_32_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.neuroimage.2009.10.002"},{"key":"e_1_2_8_33_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.neuroimage.2020.117373"},{"key":"e_1_2_8_34_1","doi-asserted-by":"publisher","DOI":"10.3389\/fnana.2016.00112"},{"key":"e_1_2_8_35_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41531\u2010024\u201000631\u20103"},{"key":"e_1_2_8_36_1","doi-asserted-by":"crossref","unstructured":"ZamirSW AroraA KhanS HayatM KhanFS YangM\u2010H Restormer: efficient transformer for high\u2010resolution image restoration.arXiv. Preprint posted online 2022.","DOI":"10.1109\/CVPR52688.2022.00564"},{"key":"e_1_2_8_37_1","doi-asserted-by":"publisher","DOI":"10.1109\/TMI.2022.3167808"},{"key":"e_1_2_8_38_1","doi-asserted-by":"crossref","unstructured":"YueZ WangJ LoyCC ResShift: Efficient Diffusion Model for Image Super\u2010resolution by Residual Shifting.arXiv. Preprint posted online 2023.","DOI":"10.52202\/075280-0583"},{"key":"e_1_2_8_39_1","doi-asserted-by":"publisher","DOI":"10.1002\/mrm.30290"},{"key":"e_1_2_8_40_1","doi-asserted-by":"crossref","unstructured":"LiL SuC GuoY ZhangH LiangD ShangK Interactive gadolinium\u2010free MRI synthesis: a transformer with localization prompt learning.arXiv [eessIV]. Preprint posted online 2025.","DOI":"10.1007\/978-3-032-31663-9_28"},{"key":"e_1_2_8_41_1","doi-asserted-by":"crossref","unstructured":"ShiW CaballeroJ Husz\u00e1rF et\u00a0al.Real\u2010time single image and video super\u2010resolution using an efficient sub\u2010pixel convolutional neural network.arXiv [csCV].Preprint posted online 2016.","DOI":"10.1109\/CVPR.2016.207"},{"key":"e_1_2_8_42_1","unstructured":"HendrycksD GimpelK Gaussian error linear units (GELUs).arXiv [csLG]. Preprint posted online 2023."},{"key":"e_1_2_8_43_1","doi-asserted-by":"publisher","DOI":"10.1109\/TIP.2003.819861"},{"key":"e_1_2_8_44_1","doi-asserted-by":"publisher","DOI":"10.2105\/AJPH.86.5.726"},{"key":"e_1_2_8_45_1","doi-asserted-by":"publisher","DOI":"10.5334\/irsp.82"},{"key":"e_1_2_8_46_1","doi-asserted-by":"publisher","DOI":"10.1109\/42.906424"},{"key":"e_1_2_8_47_1","doi-asserted-by":"crossref","unstructured":"CuiQ TosunD MukherjeeP Abbasi\u2010AslR 7T MRI Synthesization from 3T Acquisitions.Proceedings of 27th International Conference on Medical Image Computing and Computer Assisted Intervention Morocco 6\u201010 October 2024.MICCAI;2024","DOI":"10.1007\/978-3-031-72104-5_4"},{"key":"e_1_2_8_48_1","doi-asserted-by":"publisher","DOI":"10.1002\/jmri.28944"},{"key":"e_1_2_8_49_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2023.102806"},{"key":"e_1_2_8_50_1","doi-asserted-by":"crossref","unstructured":"ZhaiX KolesnikovA HoulsbyN BeyerL Scaling vision transformers.ArXiv. Preprint posted online 2022.","DOI":"10.1109\/CVPR52688.2022.01179"},{"key":"e_1_2_8_51_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41597-024-03073-x"},{"key":"e_1_2_8_52_1","doi-asserted-by":"crossref","unstructured":"TangY YangD LiW et\u00a0al.Self\u2010supervised pre\u2010training of swin transformers for 3d medical image analysis.ArXiv. Preprint posted online 2022.","DOI":"10.1109\/CVPR52688.2022.02007"},{"key":"e_1_2_8_53_1","doi-asserted-by":"publisher","DOI":"10.2967\/jnumed.121.262464"},{"key":"e_1_2_8_54_1","doi-asserted-by":"publisher","DOI":"10.1088\/2057\u20101976\/ac3055"}],"container-title":["Medical Physics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/aapm.onlinelibrary.wiley.com\/doi\/pdf\/10.1002\/mp.70496","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/aapm.onlinelibrary.wiley.com\/doi\/full-xml\/10.1002\/mp.70496","content-type":"application\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/aapm.onlinelibrary.wiley.com\/doi\/pdf\/10.1002\/mp.70496","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,15]],"date-time":"2026-09-15T16:43:56Z","timestamp":1789490636000},"score":1,"resource":{"primary":{"URL":"https:\/\/aapm.onlinelibrary.wiley.com\/doi\/10.1002\/mp.70496"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,5]]},"references-count":53,"journal-issue":{"issue":"5","published-print":{"date-parts":[[2026,5]]}},"alternative-id":["10.1002\/mp.70496"],"URL":"https:\/\/doi.org\/10.1002\/mp.70496","archive":["Portico"],"relation":{},"ISSN":["0094-2405","2473-4209"],"issn-type":[{"value":"0094-2405","type":"print"},{"value":"2473-4209","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,5]]},"assertion":[{"value":"2025-12-05","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-04-19","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-05-21","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}],"article-number":"e70496"}}