{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,30]],"date-time":"2026-10-30T18:18:14Z","timestamp":1793384294678,"version":"4.5.0"},"reference-count":36,"publisher":"MIT Press","license":[{"start":{"date-parts":[[2024,5,28]],"date-time":"2024-05-28T00:00:00Z","timestamp":1716854400000},"content-version":"vor","delay-in-days":148,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["direct.mit.edu"],"crossmark-restriction":true},"short-container-title":[],"published-print":{"date-parts":[[2024,6,13]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:p>We use generative AI to enable rapid diffusion MRI (dMRI) with high fidelity, reproducibility, and generalizability across clinical and research settings. We employ a Swin UNEt Transformers (SWIN) model, trained on Human Connectome Project (HCP) data (n = 1021) and conditioned on registered T1 scans, to perform generalized dMRI denoising. We also qualitatively demonstrate super-resolution with artificially downsampled HCP data. Remarkably, SWIN can be fine-tuned for an out-of-domain dataset with a single example scan, as we demonstrate on dMRI of children with neurodevelopmental disorders (n = 40), adults with acute traumatic brain injury (n = 40), and adolescents with intracerebral hemorrhage due to vascular malformations undergoing resection (n = 8), each cohort scanned on different scanner models with different imaging protocols at different sites. This robustness to scan acquisition parameters, patient populations, scanner types, and sites eliminates the advantages of self-supervised methods over our fully supervised generative AI approach. We exceed current state-of-the-art denoising methods in accuracy and test\u2013retest reliability of rapid diffusion tensor imaging (DTI) requiring only 90 seconds of scan time. SWIN denoising also achieves dramatic improvements over the state-of-the-art for test\u2013retest reliability of intracellular volume fraction and free water fraction measurements and can remove heavy-tail noise, improving biophysical modeling fidelity. SWIN enables rapid diffusion MRI with unprecedented accuracy and reliability, especially at high diffusion weighting for probing biological tissues at microscopic spatial scales. The code and model are publicly available at https:\/\/github.com\/ucsfncl\/dmri-swin.<\/jats:p>","DOI":"10.1162\/imag_a_00193","type":"journal-article","created":{"date-parts":[[2024,6,10]],"date-time":"2024-06-10T07:25:59Z","timestamp":1718004359000},"update-policy":"https:\/\/doi.org\/10.1162\/mitpressjournals.corrections.policy","source":"Crossref","is-referenced-by-count":6,"title":["Generative AI for rapid diffusion MRI with improved image quality, reliability, and generalizability"],"prefix":"10.1162","volume":"2","author":[{"given":"Amir","family":"Sadikov","sequence":"first","affiliation":[{"name":"Radiology and Biomedical Imaging, University of California, San Francisco, CA, United States"},{"name":"Graduate Group in Bioengineering, University of California, San Francisco, CA, United States"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xinlei","family":"Pan","sequence":"additional","affiliation":[{"name":"University of California, Berkeley, CA, United States"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Hannah","family":"Choi","sequence":"additional","affiliation":[{"name":"Radiology and Biomedical Imaging, University of California, San Francisco, CA, United States"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Lanya T.","family":"Cai","sequence":"additional","affiliation":[{"name":"Radiology and Biomedical Imaging, University of California, San Francisco, CA, United States"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Pratik","family":"Mukherjee","sequence":"additional","affiliation":[{"name":"Radiology and Biomedical Imaging, University of California, San Francisco, CA, United States"},{"name":"Graduate Group in Bioengineering, University of California, San Francisco, CA, United States"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"281","published-online":{"date-parts":[[2024,6,13]]},"reference":[{"key":"2025081117031115900_b1","doi-asserted-by":"crossref","first-page":"1063","DOI":"10.1016\/j.neuroimage.2015.10.019","article-title":"An integrated approach to correction for off-resonance effects and subject movement in diffusion MR imaging","volume":"125","author":"Andersson","year":"2016","journal-title":"NeuroImage"},{"key":"2025081117031115900_b2","first-page":"538","article-title":"Robust segmentation of brain MRI in the wild with hierarchical CNNs and no retraining","volume":"13435","author":"Billot","year":"2022","journal-title":"Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)"},{"key":"2025081117031115900_b3","article-title":"MONAI: An open-source framework for deep learning in healthcare","author":"Cardoso","year":"2022","journal-title":"ArXiv"},{"key":"2025081117031115900_b4","doi-asserted-by":"crossref","first-page":"1194","DOI":"10.1002\/jmri.22535","article-title":"Quality assessment of high angular resolution diffusion imaging data using bootstrap on q-ball reconstruction","volume":"33","author":"Cohen-Adad","year":"2011","journal-title":"Journal of Magnetic Resonance Imaging: JMRI"},{"key":"2025081117031115900_b5","unstructured":"Dehghani, M., Mustafa, B., Djolonga, J., Heek, J., Minderer, M., Caron, M., Steiner, A., Puigcerver, J., Geirhos, R., Alabdulmohsin, I., Oliver, A., Padlewski, P., Gritsenko, A., Lu\u010di\u0107, M., & Houlsby, N. (2023). Patch n\u2019 pack: NaViT, a vision transformer for any aspect ratio and resolution. ArXiv. https:\/\/doi.org\/10.48550\/arXiv.2307.06304"},{"key":"2025081117031115900_b6","article-title":"An image is worth 16x16 words: Transformers for image recognition at scale","author":"Dosovitskiy","year":"2020","journal-title":"ArXiv"},{"key":"2025081117031115900_b7","doi-asserted-by":"crossref","first-page":"198","DOI":"10.1109\/WACV56688.2023.00028","article-title":"Lossy image compression with quantized hierarchical VAEs","volume-title":"2023 IEEE\/CVF Winter Conference on Applications of Computer Vision (WACV)","author":"Duan","year":"2023"},{"key":"2025081117031115900_b8","doi-asserted-by":"crossref","first-page":"62","DOI":"10.1016\/j.neuroimage.2013.05.041","article-title":"The WU-Minn human connectome project: An overview","volume":"80","author":"Essen","year":"2013","journal-title":"NeuroImage"},{"key":"2025081117031115900_b9","article-title":"Patch2Self: Denoising diffusion MRI with self-supervised learning","volume":"2020-Decem","author":"Fadnavis","year":"2020","journal-title":"Advances in Neural Information Processing Systems"},{"key":"2025081117031115900_b10","doi-asserted-by":"crossref","first-page":"63","DOI":"10.1016\/j.neuroimage.2009.06.060","article-title":"Accurate and robust brain image alignment using boundary-based registration","volume":"48","author":"Greve","year":"2009","journal-title":"NeuroImage"},{"key":"2025081117031115900_b11","first-page":"272","article-title":"Swin UNETR: Swin transformers for semantic segmentation of brain tumors in MRI images","volume":"12962","author":"Hatamizadeh","year":"2022","journal-title":"Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)"},{"key":"2025081117031115900_b12","doi-asserted-by":"crossref","first-page":"119474","DOI":"10.1016\/j.neuroimage.2022.119474","article-title":"SynthStrip: Skull-stripping for any brain image","volume":"260","author":"Hoopes","year":"2022","journal-title":"NeuroImage"},{"key":"2025081117031115900_b13","doi-asserted-by":"crossref","first-page":"807","DOI":"10.1002\/mrm.20033","article-title":"The effect of gradient sampling schemes on measures derived from diffusion tensor MRI: A Monte Carlo study","volume":"51","author":"Jones","year":"2004","journal-title":"Magnetic Resonance in Medicine"},{"key":"2025081117031115900_b14","doi-asserted-by":"crossref","DOI":"10.1093\/med\/9780195369779.001.0001","volume-title":"Diffusion MRI","author":"Jones","year":"2012"},{"key":"2025081117031115900_b15","doi-asserted-by":"crossref","first-page":"102330","DOI":"10.1016\/j.artmed.2022.102330","article-title":"Diffusion tensor estimation with transformer neural networks","volume":"130","author":"Karimi","year":"2022","journal-title":"Artificial Intelligence in Medicine"},{"key":"2025081117031115900_b16","first-page":"371","article-title":"Left-ventricle quantification using residual U-net","volume-title":"Statistical Atlases and Computational Models of the Heart. Atrial Segmentation and LV Quantification Challenges. STACOM 2018","author":"Kerfoot","year":"2019"},{"key":"2025081117031115900_b17","doi-asserted-by":"crossref","first-page":"4441","DOI":"10.1002\/hbm.24713","article-title":"Longitudinal increases in structural connectome segregation and functional connectome integration are associated with better recovery after mild tbi","volume":"40","author":"Kuceyeski","year":"2019","journal-title":"Human Brain Mapping"},{"key":"2025081117031115900_b18","doi-asserted-by":"crossref","first-page":"1431","DOI":"10.1002\/hbm.21292","article-title":"Independent component analysis of DTI reveals multivariate microstructural correlations of white matter in the human brain","volume":"33","author":"Li","year":"2012","journal-title":"Human Brain Mapping"},{"key":"2025081117031115900_b19","first-page":"9992","article-title":"Swin transformer: Hierarchical vision transformer using shifted windows","volume-title":"Proceedings of the IEEE International Conference on Computer Vision","author":"Liu","year":"2021"},{"key":"2025081117031115900_b20","article-title":"Decoupled weight decay regularization","volume-title":"7th International Conference on Learning Representations, ICLR 2019","author":"Loshchilov","year":"2017"},{"key":"2025081117031115900_b21","doi-asserted-by":"crossref","first-page":"615","DOI":"10.1109\/TCI.2020.2964201","article-title":"MRI super-resolution with ensemble learning and complementary priors","volume":"6","author":"Lyu","year":"2019","journal-title":"IEEE Transactions on Computational Imaging"},{"key":"2025081117031115900_b22","doi-asserted-by":"crossref","first-page":"119","DOI":"10.1109\/TIP.2012.2210725","article-title":"Nonlocal transform-domain filter for volumetric data denoising and reconstruction","volume":"22","author":"Maggioni","year":"2013","journal-title":"IEEE Transactions on Image Processing"},{"key":"2025081117031115900_b23","doi-asserted-by":"crossref","first-page":"1136424","DOI":"10.3389\/fnins.2023.1136424","article-title":"Neurite orientation dispersion and density imaging of white matter microstructure in sensory processing dysfunction with versus without comorbid ADHD","volume":"17","author":"Mark","year":"2023","journal-title":"Frontiers in Neuroscience"},{"key":"2025081117031115900_b24","doi-asserted-by":"crossref","first-page":"413","DOI":"10.1002\/mrm.28395","article-title":"Training a neural network for Gibbs and noise removal in diffusion MRI","volume":"85","author":"Muckley","year":"2021","journal-title":"Magnetic Resonance in Medicine"},{"key":"2025081117031115900_b25","doi-asserted-by":"crossref","first-page":"843","DOI":"10.3174\/ajnr.A1052","article-title":"Diffusion tensor MR imaging and fiber tractography: Technical considerations","volume":"29","author":"Mukherjee","year":"2008","journal-title":"American Journal of Neuroradiology"},{"key":"2025081117031115900_b26","doi-asserted-by":"crossref","first-page":"1160","DOI":"10.1002\/mrm.29478","article-title":"Tensor denoising of multidimensional \u201cMRI\u201d data","volume":"89","author":"Olesen","year":"2023","journal-title":"Magnetic Resonance in Medicine"},{"key":"2025081117031115900_b27","article-title":"PyTorch: An imperative style, high-performance deep learning library","author":"Paszke","year":"2019","journal-title":"ArXiv"},{"key":"2025081117031115900_b28","article-title":"Grokking: Generalization beyond overfitting on small algorithmic datasets","author":"Power","year":"2022","journal-title":"ArXiv"},{"key":"2025081117031115900_b29","doi-asserted-by":"crossref","first-page":"340","DOI":"10.1006\/jmre.2000.2209","article-title":"Condition number as a measure of noise performance of diffusion tensor data acquisition schemes with MRI","volume":"147","author":"Skare","year":"2000","journal-title":"Journal of Magnetic Resonance"},{"key":"2025081117031115900_b30","doi-asserted-by":"crossref","first-page":"1487","DOI":"10.1016\/j.neuroimage.2006.02.024","article-title":"Tract-based spatial statistics: Voxelwise analysis of multi-subject diffusion data","volume":"31","author":"Smith","year":"2006","journal-title":"NeuroImage"},{"key":"2025081117031115900_b31","doi-asserted-by":"crossref","first-page":"117017","DOI":"10.1016\/j.neuroimage.2020.117017","article-title":"DeepDTI: High-fidelity six-direction diffusion tensor imaging using deep learning","volume":"219","author":"Tian","year":"2020","journal-title":"NeuroImage"},{"key":"2025081117031115900_b32","doi-asserted-by":"crossref","first-page":"119033","DOI":"10.1016\/j.neuroimage.2022.119033","article-title":"SDnDTI: Self-supervised deep learning-based denoising for diffusion tensor MRI","volume":"253","author":"Tian","year":"2022","journal-title":"NeuroImage"},{"key":"2025081117031115900_b33","doi-asserted-by":"crossref","first-page":"394","DOI":"10.1016\/j.neuroimage.2016.08.016","article-title":"Denoising of diffusion MRI using random matrix theory","volume":"142","author":"Veraart","year":"2016","journal-title":"NeuroImage"},{"key":"2025081117031115900_b34","doi-asserted-by":"crossref","first-page":"531","DOI":"10.1016\/j.neuroimage.2010.02.072","article-title":"Microstructural correlations of white matter tracts in the human brain","volume":"51","author":"Wahl","year":"2010","journal-title":"NeuroImage"},{"key":"2025081117031115900_b35","article-title":"DDM2: Self-supervised diffusion MRI denoising with generative diffusion models","author":"Xiang","year":"2023"},{"key":"2025081117031115900_b36","doi-asserted-by":"crossref","first-page":"1000","DOI":"10.1016\/j.neuroimage.2012.03.072","article-title":"NODDI: Practical in vivo neurite orientation dispersion and density imaging of the human brain","volume":"61","author":"Zhang","year":"2012","journal-title":"NeuroImage"}],"container-title":["Imaging Neuroscience"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/direct.mit.edu\/imag\/article-pdf\/doi\/10.1162\/imag_a_00193\/2383089\/imag_a_00193.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/direct.mit.edu\/imag\/article-pdf\/doi\/10.1162\/imag_a_00193\/2383089\/imag_a_00193.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,8,11]],"date-time":"2025-08-11T21:03:20Z","timestamp":1754946200000},"score":1,"resource":{"primary":{"URL":"https:\/\/direct.mit.edu\/imag\/article\/doi\/10.1162\/imag_a_00193\/121200\/Generative-AI-for-rapid-diffusion-MRI-with"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024]]},"references-count":36,"URL":"https:\/\/doi.org\/10.1162\/imag_a_00193","relation":{},"ISSN":["2837-6056"],"issn-type":[{"value":"2837-6056","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2024]]},"published":{"date-parts":[[2024]]},"article-number":"imag-2-00193"}}