{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,13]],"date-time":"2026-08-13T13:32:09Z","timestamp":1786627929980,"version":"build-2736575974"},"reference-count":26,"publisher":"Wiley","license":[{"start":{"date-parts":[[2026,8,13]],"date-time":"2026-08-13T00:00:00Z","timestamp":1786579200000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/onlinelibrary.wiley.com\/termsAndConditions#vor"},{"start":{"date-parts":[[2026,8,13]],"date-time":"2026-08-13T00:00:00Z","timestamp":1786579200000},"content-version":"tdm","delay-in-days":0,"URL":"http:\/\/doi.wiley.com\/10.1002\/tdm_license_1.1"}],"content-domain":{"domain":["onlinelibrary.wiley.com"],"crossmark-restriction":true},"short-container-title":["Magnetic Resonance Imaging"],"abstract":"ABSTRACT<\/jats:title>\n \n Background<\/jats:title>\n Deep learning (DL)\u2010based image enhancement is widely used to improve suboptimal medical imaging. Whether it also benefits diagnostic\u2010quality MRI in downstream task performance and data\u2010efficiency remains unclear.<\/jats:p>\n <\/jats:sec>\n \n Purpose<\/jats:title>\n To investigate the impact of DL\u2010based enhancement applied to diagnostic quality structural MRI for Alzheimer's disease (AD) classification.<\/jats:p>\n <\/jats:sec>\n \n Study Type<\/jats:title>\n Retrospective.<\/jats:p>\n <\/jats:sec>\n \n Population<\/jats:title>\n \n A total of 2293 brain MRI scans from the Alzheimer's Disease Neuroimaging Initiative (ADNI) were split into training (\n n<\/jats:italic>\n \u2009=\u20091605), validation (\n n<\/jats:italic>\n \u2009=\u2009229), and internal test (\n n<\/jats:italic>\n \u2009=\u2009459) sets. Two hundred and seventy scans from the National Alzheimer's Coordinating Center (NACC) served as an external test set.\n <\/jats:p>\n <\/jats:sec>\n \n Field Strength\/Sequence<\/jats:title>\n 1.5\u2009T; 3D T1\u2010weighted gradient\u2010echo.<\/jats:p>\n <\/jats:sec>\n \n Assessment<\/jats:title>\n Each scan was enhanced by SubtleHD (SHD), an FDA\u2010cleared DL\u2010based MR enhancement tool. ResNet34 and DenseNet121 were trained on standard\u2010of\u2010care (SOC) and SHD\u2010enhanced images to classify subjects as cognitively normal, mild cognitive impairment, or AD and evaluated by accuracy and macro\u2013area under the receiver operating characteristic curve (macro\u2010AUC). Data efficiency was assessed by retraining on stratified training subsets (50%\u2013100%).<\/jats:p>\n <\/jats:sec>\n \n Statistical Tests<\/jats:title>\n \n McNemar test for accuracy and DeLong test for macro\u2010AUC in three\u2010class one\u2010versus\u2010rest setting (\n p<\/jats:italic>\n \u2009<\u20090.05).\n <\/jats:p>\n <\/jats:sec>\n \n Results<\/jats:title>\n \n SHD enhancement increased ResNet34 accuracy from 85.2% to 88.7% and macro\u2010AUC from 0.951 to 0.968 (both significant), and DenseNet121 accuracy from 90.2% to 92.2% (\n p<\/jats:italic>\n \u2009=\u20090.18) and macro\u2010AUC from 0.978 to 0.982 (\n p<\/jats:italic>\n \u2009=\u20090.29). Models trained on 70% of SHD\u2010enhanced dataset matched those trained on the full SOC dataset (accuracy: 85.9%, macro\u2010AUC: 0.942), indicating improved data efficiency with enhancement. In NACC, the SOC\u2010trained model achieved accuracy of 49.2% and macro\u2010AUC of 0.679 versus 63.0% and 0.819 for the SHD\u2010trained model (both significant); the SHD\u2010trained model retained an advantage on unenhanced NACC images (macro\u2010AUC: 0.772).\n <\/jats:p>\n <\/jats:sec>\n \n Data Conclusion<\/jats:title>\n DL\u2010based enhancement of diagnostic\u2010quality MRI improves downstream Alzheimer's disease classification performance and reduces the amount of training data required. This suggests that conventional definitions of image quality may underestimate the information content available for machine learning.<\/jats:p>\n <\/jats:sec>\n \n Evidence Level<\/jats:title>\n 3.<\/jats:p>\n <\/jats:sec>\n \n Technical Efficacy Stage<\/jats:title>\n 2.<\/jats:p>\n <\/jats:sec>","DOI":"10.1002\/jmri.70507","type":"journal-article","created":{"date-parts":[[2026,8,13]],"date-time":"2026-08-13T12:45:56Z","timestamp":1786625156000},"update-policy":"https:\/\/doi.org\/10.1002\/crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Deep Learning\u2010Based Enhancement of Already Diagnostic\u2010Quality\n MRI<\/scp>\n for Alzheimer's Disease Classification: Effects on Model Performance and Training Data Requirements"],"prefix":"10.1002","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-1805-7705","authenticated-orcid":false,"given":"Zheng","family":"Zhang","sequence":"first","affiliation":[{"name":"Subtle Medical Inc. Menlo Park California USA"},{"name":"University of Surrey Surrey UK"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8149-7546","authenticated-orcid":false,"given":"Zechen","family":"Zhou","sequence":"additional","affiliation":[{"name":"Subtle Medical Inc. Menlo Park California USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Lei","family":"Xiang","sequence":"additional","affiliation":[{"name":"Subtle Medical Inc. Menlo Park California USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9395-8768","authenticated-orcid":false,"given":"Ajit","family":"Shankaranarayanan","sequence":"additional","affiliation":[{"name":"Subtle Medical Inc. Menlo Park California USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Enhao","family":"Gong","sequence":"additional","affiliation":[{"name":"Subtle Medical Inc. Menlo Park California USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5781-8848","authenticated-orcid":false,"given":"Greg","family":"Zaharchuk","sequence":"additional","affiliation":[{"name":"Stanford University Stanford California USA"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"311","published-online":{"date-parts":[[2026,8,13]]},"reference":[{"issue":"7","key":"e_1_2_11_2_1","doi-asserted-by":"crossref","first-page":"3142","DOI":"10.1109\/TIP.2017.2662206","article-title":"Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising","volume":"26","author":"Zhang K.","year":"2017","journal-title":"IEEE Transactions on Image Processing"},{"issue":"12","key":"e_1_2_11_3_1","doi-asserted-by":"crossref","first-page":"2524","DOI":"10.1109\/TMI.2017.2715284","article-title":"Low\u2010Dose CT With a Residual Encoder\u2010Decoder Convolutional Neural Network","volume":"36","author":"Chen H.","year":"2017","journal-title":"IEEE Transactions on Medical Imaging"},{"issue":"6","key":"e_1_2_11_4_1","doi-asserted-by":"crossref","first-page":"3055","DOI":"10.1002\/mrm.26977","article-title":"Learning a Variational Network for Reconstruction of Accelerated MRI Data","volume":"79","author":"Hammernik K.","year":"2018","journal-title":"Magnetic Resonance in Medicine"},{"issue":"7697","key":"e_1_2_11_5_1","doi-asserted-by":"crossref","first-page":"487","DOI":"10.1038\/nature25988","article-title":"Image Reconstruction by Domain\u2010Transform Manifold Learning","volume":"555","author":"Zhu B.","year":"2018","journal-title":"Nature"},{"issue":"2","key":"e_1_2_11_6_1","doi-asserted-by":"crossref","DOI":"10.1148\/radiol.243819","article-title":"The Current Status of AI\u2010Accelerated MRI Techniques in Clinical Use","volume":"317","author":"Haller S.","year":"2025","journal-title":"Radiology"},{"issue":"1","key":"e_1_2_11_7_1","doi-asserted-by":"crossref","first-page":"197","DOI":"10.1007\/s00062-021-01121-2","article-title":"Deep Learning Image Processing Enables 40% Faster Spinal MR Scans Which Match or Exceed Quality of Standard of Care: A Prospective Multicenter Multireader Study","volume":"32","author":"Bash S.","year":"2022","journal-title":"Clinical Neuroradiology"},{"issue":"2","key":"e_1_2_11_8_1","doi-asserted-by":"crossref","DOI":"10.1148\/ryai.210059","article-title":"Clinical Assessment of Deep Learning\u2013Based Super\u2010Resolution for 3D Volumetric Brain MRI","volume":"4","author":"Rudie J. D.","year":"2022","journal-title":"Radiology Artificial Intelligence"},{"key":"e_1_2_11_9_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2019.05.001"},{"issue":"5","key":"e_1_2_11_10_1","doi-asserted-by":"crossref","first-page":"1257","DOI":"10.1002\/jmri.70221","article-title":"Alzheimer's Disease Neuroimaging Initiative. Evaluation of Image\u2010Level Harmonization Methods for Multi\u2010Center MR Neuroimaging","volume":"63","author":"Ho B. C.","year":"2026","journal-title":"Journal of Magnetic Resonance Imaging"},{"key":"e_1_2_11_11_1","doi-asserted-by":"crossref","DOI":"10.1016\/j.cmpbup.2025.100220","article-title":"A Review and Systematic Guide to Counteracting Medical Data Scarcity for AI Applications","volume":"8","author":"Gr\u00f6ger F.","year":"2025","journal-title":"Computer Methods and Programs in Biomedicine Update"},{"key":"e_1_2_11_12_1","first-page":"506","volume-title":"Carbon Footprint of Selecting and Training Deep Learning Models for Medical Image Analysis","author":"Selvan R.","year":"2022"},{"key":"e_1_2_11_13_1","doi-asserted-by":"crossref","first-page":"259","DOI":"10.1186\/s12880-025-01775-1","article-title":"Preoperative MRI\u2010Based Deep Learning Reconstruction and Classification Model for Assessing Rectal Cancer","volume":"25","author":"Yuan Y.","year":"2025","journal-title":"BMC Medical Imaging"},{"issue":"1","key":"e_1_2_11_14_1","doi-asserted-by":"crossref","first-page":"192","DOI":"10.1038\/s41746-023-00943-3","article-title":"DeepQuality Improves Infant Retinopathy Screening","volume":"6","author":"Li L.","year":"2023","journal-title":"npj Digital Medicine"},{"issue":"2","key":"e_1_2_11_15_1","doi-asserted-by":"crossref","first-page":"928","DOI":"10.1007\/s00330-025-11920-7","article-title":"Deep Learning\u2010Based Image Enhancement for Improved Black Blood Imaging in Brain Metastasis","volume":"36","author":"Oh G.","year":"2026","journal-title":"European Radiology"},{"issue":"10","key":"e_1_2_11_16_1","doi-asserted-by":"crossref","first-page":"4826","DOI":"10.1109\/JBHI.2022.3168604","article-title":"Deep Perceptual Enhancement for Medical Image Analysis","volume":"26","author":"Sharif S. M.","year":"2022","journal-title":"IEEE Journal of Biomedical and Health Informatics"},{"issue":"1","key":"e_1_2_11_17_1","doi-asserted-by":"crossref","first-page":"89","DOI":"10.1007\/s12021-023-09646-2","article-title":"A Deep Learning\u2010Based Ensemble Method for Early Diagnosis of Alzheimer's Disease Using MRI Images","volume":"22","author":"Fathi S.","year":"2024","journal-title":"Neuroinformatics"},{"key":"e_1_2_11_18_1","doi-asserted-by":"crossref","first-page":"57","DOI":"10.1007\/s12021-020-09469-5","article-title":"Reproducible Evaluation of Diffusion MRI Features for Automatic Classification of Patients With Alzheimer's Disease","volume":"19","author":"Wen J.","year":"2021","journal-title":"Neuroinformatics"},{"issue":"1","key":"e_1_2_11_19_1","doi-asserted-by":"crossref","first-page":"49","DOI":"10.3174\/ajnr.A8424","article-title":"DANTE\u2010CAIPI Accelerated Contrast\u2010Enhanced 3D T1: Deep Learning\u2013Based Image Quality Improvement for Vessel Wall MRI","volume":"46","author":"Kharaji M.","year":"2025","journal-title":"American Journal of Neuroradiology"},{"issue":"12","key":"e_1_2_11_20_1","doi-asserted-by":"crossref","first-page":"2130","DOI":"10.3174\/ajnr.A7358","article-title":"Deep Learning Enables 60% Accelerated Volumetric Brain MRI While Preserving Quantitative Performance: A Prospective, Multicenter, Multireader Trial","volume":"42","author":"Bash S.","year":"2021","journal-title":"American Journal of Neuroradiology"},{"key":"e_1_2_11_21_1","first-page":"770","volume-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition","author":"He K.","year":"2016"},{"key":"e_1_2_11_22_1","first-page":"4700","volume-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition","author":"Huang G.","year":"2017"},{"issue":"1","key":"e_1_2_11_23_1","doi-asserted-by":"crossref","DOI":"10.1186\/s41747-025-00562-5","article-title":"AI for Image Quality and Patient Safety in CT and MRI","volume":"9","author":"Melazzini L.","year":"2025","journal-title":"European Radiology Experimental"},{"issue":"4","key":"e_1_2_11_24_1","doi-asserted-by":"crossref","first-page":"685","DOI":"10.1002\/jmri.21049","article-title":"The Alzheimer's Disease Neuroimaging Initiative (ADNI): MRI Methods","volume":"27","author":"Jack C. R.","year":"2008","journal-title":"Journal of Magnetic Resonance Imaging"},{"issue":"6","key":"e_1_2_11_25_1","doi-asserted-by":"crossref","first-page":"e271","DOI":"10.1016\/S2589-7500(19)30123-2","article-title":"A Comparison of Deep Learning Performance Against Health\u2010Care Professionals in Detecting Diseases From Medical Imaging: A Systematic Review and Meta\u2010Analysis","volume":"1","author":"Liu X.","year":"2019","journal-title":"Lancet Digital Health"},{"key":"e_1_2_11_26_1","article-title":"Artificial Intelligence Versus Clinicians: Systematic Review of Design, Reporting Standards, and Claims of Deep Learning Studies","volume":"689","author":"Nagendran M.","year":"2020","journal-title":"BMJ"},{"issue":"9","key":"e_1_2_11_27_1","doi-asserted-by":"crossref","first-page":"2613","DOI":"10.1038\/s41591-024-03097-1","article-title":"Evaluation and Mitigation of the Limitations of Large Language Models in Clinical Decision\u2010Making","volume":"30","author":"Hager P.","year":"2024","journal-title":"Nature Medicine"}],"container-title":["Journal of Magnetic Resonance Imaging"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/pdf\/10.1002\/jmri.70507","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/full-xml\/10.1002\/jmri.70507","content-type":"application\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/pdf\/10.1002\/jmri.70507","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,8,13]],"date-time":"2026-08-13T12:46:00Z","timestamp":1786625160000},"score":1,"resource":{"primary":{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/10.1002\/jmri.70507"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8,13]]},"references-count":26,"alternative-id":["10.1002\/jmri.70507"],"URL":"https:\/\/doi.org\/10.1002\/jmri.70507","archive":["Portico"],"relation":{},"ISSN":["1053-1807","1522-2586"],"issn-type":[{"value":"1053-1807","type":"print"},{"value":"1522-2586","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,8,13]]},"assertion":[{"value":"2026-06-01","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-08-03","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-08-13","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}],"article-number":"jmri.70507"}}