{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,8]],"date-time":"2026-10-08T15:00:27Z","timestamp":1791471627867,"version":"4.3.4"},"reference-count":46,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,9,23]],"date-time":"2026-09-23T00:00:00Z","timestamp":1790121600000},"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. Med."],"abstract":"<jats:sec>\n                    <jats:title>Introduction<\/jats:title>\n                    <jats:p>Accurate segmentation of Multiple Sclerosis (MS) lesions from Magnetic Resonance Imaging (MRI) is fundamental for robust disease monitoring and for the evaluation of therapeutic efficacy. Within Internet of Medical Things (IoMT)-based diagnostic infrastructures, where deep learning models serve as autonomous computational nodes embedded in interconnected clinical workflows, ensuring reliable lesion detection without continuous human supervision is a critical operational and clinical requirement. A persistent methodological challenge is the pronounced class imbalance characteristic of MS neuroimaging datasets. This imbalance causes conventional loss functions to systematically miss small lesions, thereby directly compromising the utility of automated image analysis as an instrument for precision medicine.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>This work proposes the Regularization via Gradient Attribution (RGA) framework, a module that leverages gradient-based eXplainable Artificial Intelligence (XAI) attribution maps as active supervisory signals during training. Instead of constraining the learning process exclusively at the prediction-output level, RGA imposes additional penalties on the network when false-negative lesion regions have low internal attribution and when false-positive healthy-tissue regions exhibit elevated attribution. The framework is architecturally agnostic, compatible with both convolutional and transformer-based backbone networks, and accommodating arbitrary spatial attribution techniques. RGA was evaluated on two public MS lesion segmentation datasets, MSLesSeg and 3D-MR-MS, using three backbones (nnU-Net, UNETR, TransBTS) and two XAI methods (LayerCAM, Integrated Gradients) in 5-fold cross-validation.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>On both datasets, RGA consistently improved DSC, TPR, and LTPR over the baselines. The best configuration, nnU-Net + RGA achieved a DSC of 0.7056 on MSLesSeg and 0.7466 on 3D-MR-MS, with RGA yielding the highest DSC for all backbones. Qualitative analyses further substantiate these findings by demonstrating a marked reduction in missed-lesion regions across all model architectures, without a corresponding substantial increase in false-positive predictions.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Discussion<\/jats:title>\n                    <jats:p>By jointly providing high lesion-level detection completeness, model interpretability, and zero additional inference-time computational cost, RGA represents a methodologically grounded approach compatible with IoMT-oriented precision-medicine pipelines, while requiring dedicated deployment validation before clinical or edge-device implementation.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/fmed.2026.1907567","type":"journal-article","created":{"date-parts":[[2026,9,23]],"date-time":"2026-09-23T04:49:23Z","timestamp":1790138963000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Regularization via gradient attribution for multiple sclerosis lesion segmentation"],"prefix":"10.3389","volume":"13","author":[{"given":"Salvatore","family":"Giugliano","sequence":"first","affiliation":[{"name":"Institute for High-Performance Computing and Networking (ICAR), National Research Council of Italy (CNR)","place":["Naples, Italy"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Giovanna","family":"Sannino","sequence":"additional","affiliation":[{"name":"Institute for High-Performance Computing and Networking (ICAR), National Research Council of Italy (CNR)","place":["Naples, Italy"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,9,23]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","first-page":"1858","DOI":"10.1093\/brain\/awz144","article-title":"Assessment of lesions on magnetic resonance imaging in multiple sclerosis: practical guidelines","volume":"142","author":"Filippi","year":"2019","journal-title":"Brain"},{"key":"B2","doi-asserted-by":"publisher","first-page":"156","DOI":"10.4258\/hir.2016.22.3.156","article-title":"Medical internet of things and big data in healthcare","volume":"22","author":"Dimitrov","year":"2016","journal-title":"Healthc Inform Res"},{"key":"B3","doi-asserted-by":"publisher","first-page":"597","DOI":"10.1038\/nrneurol.2015.157","article-title":"Evidence-based guidelines: magnims consensus guidelines on the use of MRI in multiple sclerosis-establishing disease prognosis and monitoring patients","volume":"11","author":"Wattjes","year":"2015","journal-title":"Nat Rev Neurol"},{"key":"B4","doi-asserted-by":"publisher","first-page":"669","DOI":"10.1016\/S1474-4422(13)70103-0","article-title":"MRI lesions as a surrogate for relapses in multiple sclerosis: a meta-analysis of randomised trials","volume":"12","author":"Sormani","year":"2013","journal-title":"Lancet Neurol"},{"key":"B5","doi-asserted-by":"publisher","first-page":"264","DOI":"10.1016\/j.nicl.2016.11.020","article-title":"MRI flair lesion segmentation in multiple sclerosis: does automated segmentation hold up with manual annotation?","volume":"13","author":"Egger","year":"2017","journal-title":"NeuroImage Clin"},{"key":"B6","doi-asserted-by":"publisher","first-page":"610967","DOI":"10.3389\/fninf.2020.610967","article-title":"Review of deep learning approaches for the segmentation of multiple sclerosis lesions on brain MRI","volume":"14","author":"Zeng","year":"2020","journal-title":"Front Neuroinform"},{"key":"B7","doi-asserted-by":"publisher","first-page":"118589","DOI":"10.1016\/j.neuroimage.2021.118589","article-title":"Multiple sclerosis lesions segmentation from multiple experts: the MICCAI 2016 challenge dataset","volume":"244","author":"Commowick","year":"2021","journal-title":"Neuroimage"},{"key":"B8","doi-asserted-by":"publisher","first-page":"1121","DOI":"10.1109\/COMST.2020.2973314","article-title":"The future of healthcare internet of things: a survey of emerging technologies","volume":"22","author":"Qadri","year":"2020","journal-title":"IEEE Commun Surv Tutorials"},{"key":"B9","doi-asserted-by":"publisher","first-page":"83","DOI":"10.1016\/j.compmedimag.2018.10.002","article-title":"Survey of automated multiple sclerosis lesion segmentation techniques on magnetic resonance imaging","volume":"70","author":"Danelakis","year":"2018","journal-title":"Comput Med Imaging Graph"},{"key":"B10","author":"Milletari","year":""},{"key":"B11","doi-asserted-by":"publisher","first-page":"19","DOI":"10.1007\/s10479-005-5724-z","article-title":"A tutorial on the cross-entropy method","volume":"134","author":"De Boer","year":"2005","journal-title":"Ann Oper Res"},{"key":"B12","author":"Sudre","year":""},{"key":"B13","doi-asserted-by":"publisher","first-page":"310","DOI":"10.1186\/s12911-020-01332-6","article-title":"Explainability for artificial intelligence in healthcare: a multidisciplinary perspective","volume":"20","author":"Amann","year":"2020","journal-title":"BMC Med Inform Decis Mak"},{"key":"B14","doi-asserted-by":"publisher","first-page":"678","DOI":"10.1109\/ACCESS.2015.2437951","article-title":"The internet of things for health care: a comprehensive survey","volume":"3","author":"Islam","year":"2015","journal-title":"IEEE Access"},{"key":"B15","doi-asserted-by":"publisher","first-page":"154","DOI":"10.1016\/j.inffus.2022.11.013","article-title":"Beyond explaining: opportunities and challenges of xai-based model improvement","volume":"92","author":"Weber","year":"2023","journal-title":"Inf Fusion"},{"key":"B16","author":"Giugliano","year":""},{"key":"B17","doi-asserted-by":"publisher","first-page":"6006","DOI":"10.1016\/j.procs.2025.10.070","article-title":"Interpretable vs. post-hoc explained models for digital twin applications in clinical forecasting: a multiple sclerosis case study","volume":"270","author":"Giugliano","year":"2025","journal-title":"Procedia Comput Sci"},{"key":"B18","doi-asserted-by":"crossref","first-page":"158","DOI":"10.1007\/s10462-026-11564-z","article-title":"Impactx: improving model performance by appropriately constraining the training with teacher explanations","volume":"59","author":"Apicella","year":"2026","journal-title":"Artif Intell Rev"},{"key":"B19","doi-asserted-by":"publisher","first-page":"105111","DOI":"10.1016\/j.compbiomed.2021.105111","article-title":"Transparency of deep neural networks for medical image analysis: a review of interpretability methods","volume":"140","author":"Salahuddin","year":"2022","journal-title":"Comput Biol Med"},{"key":"B20","author":"Dosovitskiy","year":""},{"key":"B21","doi-asserted-by":"publisher","first-page":"107112","DOI":"10.1109\/ACCESS.2020.3000322","article-title":"An effective training scheme for deep neural network in edge computing enabled internet of medical things (IoMT) systems","volume":"8","author":"Pustokhina","year":"2020","journal-title":"IEEE Access"},{"key":"B22","doi-asserted-by":"publisher","first-page":"109202","DOI":"10.1016\/j.compeleceng.2024.109202","article-title":"Edge AI for internet of medical things: a literature review","volume":"116","author":"Rocha","year":"2024","journal-title":"Comput Electr Eng"},{"key":"B23","doi-asserted-by":"publisher","first-page":"44","DOI":"10.1038\/s41591-018-0300-7","article-title":"High-performance medicine: the convergence of human and artificial intelligence","volume":"25","author":"Topol","year":"2019","journal-title":"Nat Med"},{"key":"B24","author":"\u00c7i\u00e7ek","year":""},{"key":"B25","doi-asserted-by":"publisher","first-page":"203","DOI":"10.1038\/s41592-020-01008-z","article-title":"nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation","volume":"18","author":"Isensee","year":"2021","journal-title":"Nat Methods"},{"key":"B26","doi-asserted-by":"publisher","first-page":"257","DOI":"10.3174\/ajnr.A5012","article-title":"Synthetic MRI in the detection of multiple sclerosis plaques","volume":"38","author":"Hagiwara","year":"2017","journal-title":"Am J Neuroradiol"},{"key":"B27","author":"Roberts","year":""},{"key":"B28","doi-asserted-by":"publisher","first-page":"1645","DOI":"10.3174\/ajnr.A0645","article-title":"Improved identification of intracortical lesions in multiple sclerosis with phase-sensitive inversion recovery in combination with fast double inversion recovery MR imaging","volume":"28","author":"Nelson","year":"2007","journal-title":"Am J Neuroradiol"},{"key":"B29","doi-asserted-by":"publisher","first-page":"1586","DOI":"10.1002\/mrm.29963","article-title":"Maximum spherical mean value filtering for whole-brain qsm.[j]","volume":"91","author":"Alexandra","year":"2024","journal-title":"Magn Reson Med"},{"key":"B30","author":"Lin","year":""},{"key":"B31","author":"Salehi","year":""},{"key":"B32","author":"Abraham","year":""},{"key":"B33","author":"Hatamizadeh","year":""},{"key":"B34","author":"Wang","year":""},{"key":"B35","author":"Cao","year":""},{"key":"B36","doi-asserted-by":"publisher","first-page":"920","DOI":"10.1038\/s41597-025-05250-y","article-title":"Mslesseg: baseline and benchmarking of a new multiple sclerosis lesion segmentation dataset","volume":"12","author":"Guarnera","year":"2025","journal-title":"Sci Data"},{"key":"B37","author":"Rondinella","year":""},{"key":"B38","doi-asserted-by":"publisher","first-page":"51","DOI":"10.1007\/s12021-017-9348-7","article-title":"A novel public MR image dataset of multiple sclerosis patients with lesion segmentations based on multi-rater consensus","volume":"16","author":"Lesjak","year":"2018","journal-title":"Neuroinformatics"},{"key":"B39","doi-asserted-by":"publisher","first-page":"102940","DOI":"10.1016\/j.nicl.2022.102940","article-title":"Automatic segmentation of white matter hyperintensities: validation and comparison with state-of-the-art methods on both multiple sclerosis and elderly subjects","volume":"33","author":"Tran","year":"2022","journal-title":"NeuroImage Clin"},{"key":"B40","doi-asserted-by":"publisher","first-page":"21348","DOI":"10.1038\/s41598-024-72649-9","article-title":"Consensus of algorithms for lesion segmentation in brain MRI studies of multiple sclerosis","volume":"14","author":"De Rosa","year":"2024","journal-title":"Sci Rep"},{"key":"B41","author":"Selvaraju","year":""},{"key":"B42","doi-asserted-by":"publisher","first-page":"5875","DOI":"10.1109\/TIP.2021.3089943","article-title":"Layercam: exploring hierarchical class activation maps for localization","volume":"30","author":"Jiang","year":"2021","journal-title":"IEEE Trans Image Process"},{"key":"B43","author":"Sundararajan","year":""},{"key":"B44","author":"Muschelli","year":""},{"key":"B45","author":"Roberts","year":""},{"key":"B46","doi-asserted-by":"publisher","first-page":"106791","DOI":"10.1016\/j.compbiomed.2023.106791","article-title":"Medvit: a robust vision transformer for generalized medical image classification","volume":"157","author":"Manzari","year":"2023","journal-title":"Comput Biol Med"}],"container-title":["Frontiers in Medicine"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fmed.2026.1907567\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,10,8]],"date-time":"2026-10-08T14:42:14Z","timestamp":1791470534000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fmed.2026.1907567\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,23]]},"references-count":46,"alternative-id":["10.3389\/fmed.2026.1907567"],"URL":"https:\/\/doi.org\/10.3389\/fmed.2026.1907567","relation":{},"ISSN":["2296-858X"],"issn-type":[{"value":"2296-858X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,23]]},"article-number":"1907567"}}