{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,14]],"date-time":"2026-10-14T08:16:31Z","timestamp":1791965791163,"version":"4.3.4"},"reference-count":46,"publisher":"Public Library of Science (PLoS)","issue":"9","license":[{"start":{"date-parts":[[2021,9,1]],"date-time":"2021-09-01T00:00:00Z","timestamp":1630454400000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["www.plosone.org"],"crossmark-restriction":false},"short-container-title":["PLoS ONE"],"abstract":"<jats:p>Gadolinium-enhancing lesions reflect active disease and are critical for in-patient monitoring in multiple sclerosis (MS). In this work, we have developed the first fully automated method to segment and count the gadolinium-enhancing lesions from routine clinical MRI of MS patients. The proposed method first segments the potential lesions using 2D-UNet from multi-channel scans (T1 post-contrast, T1 pre-contrast, FLAIR, T2, and proton-density) and classifies the lesions using a random forest classifier. The algorithm was trained and validated on 600 MRIs with manual segmentation. We compared the effect of loss functions (Dice, cross entropy, and bootstrapping cross entropy) and number of input contrasts. We compared the lesion counts with those by radiologists using 2,846 images. Dice, lesion-wise sensitivity, and false discovery rate with full 5 contrasts were 0.698, 0.844, and 0.307, which improved to 0.767, 0.969, and 0.00 in large lesions (&gt;100 voxels). The model using bootstrapping loss function provided a statistically significant increase of 7.1% in sensitivity and of 2.3% in Dice compared with the model using cross entropy loss. T1 post\/pre-contrast and FLAIR were the most important contrasts. For large lesions, the 2D-UNet model trained using T1 pre-contrast, FLAIR, T2, PD had a lesion-wise sensitivity of 0.688 and false discovery rate 0.083, even without T1 post-contrast. For counting lesions in 2846 routine MRI images, the model with 2D-UNet and random forest, which was trained with bootstrapping cross entropy, achieved accuracy of 87.7% using T1 pre-contrast, T1 post-contrast, and FLAIR when lesion counts were categorized as 0, 1, and 2 or more. The model performs well in routine non-standardized MRI datasets, allows large-scale analysis of clinical datasets, and may have clinical applications.<\/jats:p>","DOI":"10.1371\/journal.pone.0255939","type":"journal-article","created":{"date-parts":[[2021,9,1]],"date-time":"2021-09-01T13:38:35Z","timestamp":1630503515000},"page":"e0255939","update-policy":"https:\/\/doi.org\/10.1371\/journal.pone.corrections_policy","source":"Crossref","is-referenced-by-count":36,"title":["Automatic segmentation of gadolinium-enhancing lesions in multiple sclerosis using deep learning from clinical MRI"],"prefix":"10.1371","volume":"16","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-6997-5717","authenticated-orcid":true,"given":"Sibaji","family":"Gaj","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Daniel","family":"Ontaneda","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Kunio","family":"Nakamura","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"340","published-online":{"date-parts":[[2021,9,1]]},"reference":[{"key":"pone.0255939.ref001","first-page":"S15","article-title":"Overview of the epidemiology, diagnosis, and disease progression associated with multiple sclerosis","volume":"19","author":"MJ Tullman","year":"2013","journal-title":"The American journal of managed care"},{"key":"pone.0255939.ref002","doi-asserted-by":"crossref","first-page":"964","DOI":"10.1016\/S0140-6736(98)03053-0","article-title":"Predictive value of gadolinium-enhanced magnetic resonance imaging for relapse rate and changes in disability or impairment in multiple sclerosis: a meta-analysis","volume":"353","author":"L Kappos","year":"1999","journal-title":"The Lancet"},{"key":"pone.0255939.ref003","doi-asserted-by":"crossref","first-page":"3","DOI":"10.1093\/brain\/121.1.3","article-title":"The role of magnetic resonance techniques in understanding and managing multiple sclerosis","volume":"121","author":"D Miller","year":"1998","journal-title":"Brain: a journal of neurology"},{"key":"pone.0255939.ref004","doi-asserted-by":"crossref","first-page":"1447","DOI":"10.1212\/01.wnl.0000183149.87975.32","article-title":"Predicting gadolinium enhancement status in MS patients eligible for randomized clinical trials","volume":"65","author":"F Barkhof","year":"2005","journal-title":"Neurology"},{"key":"pone.0255939.ref005","doi-asserted-by":"crossref","first-page":"640","DOI":"10.1212\/01.WNL.0000046587.83503.1E","article-title":"MRI contrast uptake in new lesions in relapsing-remitting MS followed at weekly intervals","volume":"60","author":"F Cotton","year":"2003","journal-title":"Neurology"},{"key":"pone.0255939.ref006","doi-asserted-by":"crossref","first-page":"22S","DOI":"10.1177\/1051228405282243","article-title":"Magnetic resonance imaging monitoring of multiple sclerosis lesion evolution","volume":"15","author":"M Inglese","year":"2005","journal-title":"Journal of Neuroimaging"},{"key":"pone.0255939.ref007","first-page":"932","article-title":"Segmentation of gadolinium-enhanced lesions on MRI in multiple sclerosis. Journal of Magnetic Resonance Imaging","volume":"25","author":"S Datta","year":"2007","journal-title":"An Official Journal of the International Society for Magnetic Resonance in Medicine"},{"key":"pone.0255939.ref008","doi-asserted-by":"crossref","first-page":"17","DOI":"10.1016\/j.media.2015.06.004","article-title":"Adaptive multi-level conditional random fields for detection and segmentation of small enhanced pathology in medical images","volume":"27","author":"Z Karimaghaloo","year":"2016","journal-title":"Medical image analysis"},{"key":"pone.0255939.ref009","doi-asserted-by":"crossref","first-page":"2663","DOI":"10.1109\/TMI.2018.2845918","article-title":"H-DenseUNet: hybrid densely connected UNet for liver and tumor segmentation from CT volumes","volume":"37","author":"X Li","year":"2018","journal-title":"IEEE transactions on medical imaging"},{"key":"pone.0255939.ref010","doi-asserted-by":"crossref","first-page":"177","DOI":"10.1148\/radiol.2018172322","article-title":"Use of 2D U-Net convolutional neural networks for automated cartilage and meniscus segmentation of knee MR imaging data to determine relaxometry and morphometry","volume":"288","author":"B Norman","year":"2018","journal-title":"Radiology"},{"key":"pone.0255939.ref011","doi-asserted-by":"crossref","first-page":"1229","DOI":"10.1109\/TMI.2016.2528821","article-title":"Deep 3D convolutional encoder networks with shortcuts for multiscale feature integration applied to multiple sclerosis lesion segmentation","volume":"35","author":"T Brosch","year":"2016","journal-title":"IEEE transactions on medical imaging"},{"key":"pone.0255939.ref012","doi-asserted-by":"crossref","first-page":"159","DOI":"10.1016\/j.neuroimage.2017.04.034","article-title":"Improving automated multiple sclerosis lesion segmentation with a cascaded 3D convolutional neural network approach","volume":"155","author":"S Valverde","year":"2017","journal-title":"NeuroImage"},{"key":"pone.0255939.ref013","doi-asserted-by":"crossref","first-page":"101638","DOI":"10.1016\/j.nicl.2018.101638","article-title":"One-shot domain adaptation in multiple sclerosis lesion segmentation using convolutional neural networks","volume":"21","author":"S Valverde","year":"2019","journal-title":"NeuroImage: Clinical"},{"key":"pone.0255939.ref014","doi-asserted-by":"crossref","unstructured":"Aslani S, Murino V, Dayan M, Tam R, Sona D and Hamarneh G. Scanner invariant multiple sclerosis lesion segmentation from MRI. arXiv preprint arXiv:191010035. 2019.","DOI":"10.1109\/ISBI45749.2020.9098721"},{"key":"pone.0255939.ref015","article-title":"Deep Learning for Predicting Enhancing Lesions in Multiple Sclerosis from Noncontrast MRI","volume":"191061","author":"PA Narayana","year":"2019","journal-title":"Radiology"},{"key":"pone.0255939.ref016","article-title":"Deep learning segmentation of gadolinium-enhancing lesions in multiple sclerosis","author":"I. Coronado","year":"2020","journal-title":"Multiple Sclerosis Journal"},{"key":"pone.0255939.ref017","unstructured":"Badrinarayanan V, Kendall A and SegNet RC. A deep convolutional encoder-decoder architecture for image segmentation. arXiv preprint arXiv:151100561. 2015; 5."},{"key":"pone.0255939.ref018","unstructured":"Jesson A and Arbel T. Hierarchical MRF and random forest segmentation of MS lesions and healthy tissues in brain MRI. Proceedings of the 2015 Longitudinal Multiple Sclerosis Lesion Segmentation Challenge. 2015: 1\u20132."},{"key":"pone.0255939.ref019","doi-asserted-by":"crossref","unstructured":"Ghafoorian M, Karssemeijer N, Heskes T, et al. Non-uniform patch sampling with deep convolutional neural networks for white matter hyperintensity segmentation. 2016 IEEE 13th International Symposium on Biomedical Imaging (ISBI). IEEE, 2016, p. 1414\u20137.","DOI":"10.1109\/ISBI.2016.7493532"},{"key":"pone.0255939.ref020","doi-asserted-by":"crossref","first-page":"378","DOI":"10.1016\/j.neuroimage.2011.03.080","article-title":"Spatial decision forests for MS lesion segmentation in multi-channel magnetic resonance images","volume":"57","author":"E Geremia","year":"2011","journal-title":"NeuroImage"},{"key":"pone.0255939.ref021","doi-asserted-by":"crossref","first-page":"87","DOI":"10.1109\/42.668698","article-title":"A nonparametric method for automatic correction of intensity nonuniformity in MRI data","volume":"17","author":"JG Sled","year":"1998","journal-title":"IEEE transactions on medical imaging"},{"key":"pone.0255939.ref022","doi-asserted-by":"crossref","first-page":"180","DOI":"10.1117\/12.185178","volume-title":"Visualization in Biomedical Computing 1994","author":"DL Collins","year":"1994"},{"key":"pone.0255939.ref023","doi-asserted-by":"crossref","first-page":"313","DOI":"10.1016\/j.neuroimage.2010.07.033","article-title":"Unbiased average age-appropriate atlases for pediatric studies","volume":"54","author":"V Fonov","year":"2011","journal-title":"Neuroimage"},{"key":"pone.0255939.ref024","doi-asserted-by":"crossref","first-page":"2362","DOI":"10.1016\/j.neuroimage.2011.09.012","article-title":"BEaST: brain extraction based on nonlocal segmentation technique","volume":"59","author":"SF Eskildsen","year":"2012","journal-title":"NeuroImage"},{"key":"pone.0255939.ref025","unstructured":"https:\/\/github.com\/BIC-MNI\/minc-toolkit-v2."},{"key":"pone.0255939.ref026","doi-asserted-by":"crossref","first-page":"2033","DOI":"10.1016\/j.neuroimage.2010.09.025","article-title":"A reproducible evaluation of ANTs similarity metric performance in brain image registration","volume":"54","author":"BB Avants","year":"2011","journal-title":"Neuroimage"},{"key":"pone.0255939.ref027","unstructured":"MINC Toolkit: http:\/\/bic-mni.github.io\/."},{"key":"pone.0255939.ref028","doi-asserted-by":"crossref","unstructured":"Ronneberger O, Fischer P and Brox T. U-net: Convolutional networks for biomedical image segmentation. International Conference on Medical image computing and computer-assisted intervention. Springer, 2015, p. 234\u201341.","DOI":"10.1007\/978-3-319-24574-4_28"},{"key":"pone.0255939.ref029","doi-asserted-by":"crossref","unstructured":"Pohlen T, Hermans A, Mathias M and Leibe B. Full-resolution residual networks for semantic segmentation in street scenes. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2017, p. 4151\u201360.","DOI":"10.1109\/CVPR.2017.353"},{"key":"pone.0255939.ref030","unstructured":"Wu Z, Shen C and Hengel Avd. Bridging category-level and instance-level semantic image segmentation. arXiv preprint arXiv:160506885. 2016."},{"key":"pone.0255939.ref031","unstructured":"Chollet F. others. 2015. Keras: Deep learning library for theano and tensorflow URL: https:\/\/keras.io\/k. 2015."},{"key":"pone.0255939.ref032","unstructured":"Abadi M, Barham P, Chen J, et al. Tensorflow: A system for large-scale machine learning. 12th {USENIX} Symposium on Operating Systems Design and Implementation ({OSDI} 16). 2016, p. 265\u201383."},{"key":"pone.0255939.ref033","unstructured":"Center OS. Ohio supercomputer center. 1987."},{"key":"pone.0255939.ref034","doi-asserted-by":"crossref","first-page":"1490","DOI":"10.1109\/TMI.2013.2258403","article-title":"Temporally consistent probabilistic detection of new multiple sclerosis lesions in brain MRI","volume":"32","author":"C Elliott","year":"2013","journal-title":"IEEE transactions on medical imaging"},{"issue":"10127","key":"pone.0255939.ref035","doi-asserted-by":"crossref","first-page":"1263","DOI":"10.1016\/S0140-6736(18)30475-6","article-title":"Siponimod versus placebo in secondary progressive multiple sclerosis (EXPAND): a double-blind, randomised, phase 3 study","volume":"391","author":"L Kappos","year":"2018","journal-title":"The Lancet"},{"issue":"2","key":"pone.0255939.ref036","doi-asserted-by":"crossref","first-page":"185","DOI":"10.1007\/s40263-020-00700-0","article-title":"Diroximel fumarate demonstrates an improved gastrointestinal tolerability profile compared with dimethyl fumarate in patients with relapsing\u2013remitting multiple sclerosis: results from the randomized, double-blind, phase III EVOLVE-MS-2 study","volume":"34","author":"RT Naismith","year":"2020","journal-title":"CNS drugs"},{"issue":"3","key":"pone.0255939.ref037","doi-asserted-by":"crossref","first-page":"221","DOI":"10.1056\/NEJMoa1601277","article-title":"Ocrelizumab versus interferon beta-1a in relapsing multiple sclerosis","volume":"376","author":"SL Hauser","year":"2017","journal-title":"New England Journal of Medicine"},{"issue":"3","key":"pone.0255939.ref038","doi-asserted-by":"crossref","first-page":"254","DOI":"10.1016\/S1474-4422(09)70021-3","article-title":"Effect of natalizumab on clinical and radiological disease activity in multiple sclerosis: a retrospective analysis of the Natalizumab Safety and Efficacy in Relapsing-Remitting Multiple Sclerosis (AFFIRM) study","volume":"8","author":"E Havrdova","year":"2009","journal-title":"The Lancet Neurology"},{"issue":"9856","key":"pone.0255939.ref039","doi-asserted-by":"crossref","first-page":"1819","DOI":"10.1016\/S0140-6736(12)61769-3","article-title":"Alemtuzumab versus interferon beta 1a as first-line treatment for patients with relapsing-remitting multiple sclerosis: a randomised controlled phase 3 trial","volume":"380","author":"JA Cohen","year":"2012","journal-title":"The Lancet"},{"issue":"9856","key":"pone.0255939.ref040","doi-asserted-by":"crossref","first-page":"1829","DOI":"10.1016\/S0140-6736(12)61768-1","article-title":"Alemtuzumab for patients with relapsing multiple sclerosis after disease-modifying therapy: a randomised controlled phase 3 trial","volume":"380","author":"AJ Coles","year":"2012","journal-title":"The lancet"},{"issue":"9","key":"pone.0255939.ref041","doi-asserted-by":"crossref","first-page":"1794","DOI":"10.1007\/s00415-014-7412-x","article-title":"Effects of delayed-release dimethyl fumarate on MRI measures in the Phase 3 DEFINE study","volume":"261","author":"DL Arnold","year":"2014","journal-title":"Journal of neurology"},{"issue":"9","key":"pone.0255939.ref042","doi-asserted-by":"crossref","first-page":"1089","DOI":"10.1001\/jamaneurol.2016.1451","article-title":"Safety and efficacy of siponimod (BAF312) in patients with relapsing-remitting multiple sclerosis: dose-blinded, randomized extension of the phase 2 BOLD study","volume":"73","author":"L Kappos","year":"2016","journal-title":"JAMA neurology"},{"key":"pone.0255939.ref043","doi-asserted-by":"crossref","first-page":"44","DOI":"10.1016\/j.msard.2016.08.002","article-title":"Comparative efficacy and discontinuation of dimethyl fumarate and fingolimod in clinical practice at 12-month follow-up","volume":"10","author":"CM Hersh","year":"2016","journal-title":"Multiple sclerosis and related disorders"},{"key":"pone.0255939.ref044","doi-asserted-by":"crossref","first-page":"1816","DOI":"10.3389\/fneur.2020.592304","article-title":"Ocrelizumab in multiple sclerosis: a real-world study from Spain","volume":"11","author":"AP Sempere","year":"2021","journal-title":"Frontiers in Neurology"},{"key":"pone.0255939.ref045","article-title":"Gadolinium Deposition in Human Brain Tissues after Contrast-enhanced MR Imaging in Adult Patients without Intracranial Abnormalities","volume":"161595","author":"RJ McDonald","year":"2017","journal-title":"Radiology"},{"issue":"3","key":"pone.0255939.ref046","doi-asserted-by":"crossref","first-page":"772","DOI":"10.1148\/radiol.15150025","article-title":"Intracranial Gadolinium Deposition after Contrast-enhanced MR Imaging","volume":"275","author":"RJ McDonald","year":"2015","journal-title":"Radiology"}],"container-title":["PLOS ONE"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0255939","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2021,9,1]],"date-time":"2021-09-01T13:39:10Z","timestamp":1630503550000},"score":1,"resource":{"primary":{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0255939"}},"subtitle":[],"editor":[{"given":"Dzung","family":"Pham","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"editor"}]}],"short-title":[],"issued":{"date-parts":[[2021,9,1]]},"references-count":46,"journal-issue":{"issue":"9","published-online":{"date-parts":[[2021,9,1]]}},"URL":"https:\/\/doi.org\/10.1371\/journal.pone.0255939","relation":{},"ISSN":["1932-6203"],"issn-type":[{"value":"1932-6203","type":"electronic"}],"subject":[],"published":{"date-parts":[[2021,9,1]]}}}