[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-406904-105":59,"doc-detail-406904-en":130},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":123,"head_meta":125,"extra_data":127,"updated_unix":129},105,"en","uniself-uniself-a-unified-network-with-instance-normalization-and-self-ensembled-lesion-fusion-for-multiple-sclerosis-lesion-segmentation","UNISELF - UNISELF - A Unified Network with Instance Normalization and Self-Ensembled Lesion Fusion for Multiple Sclerosis Lesion Segmentation","","Automated segmentation of multiple sclerosis (MS) lesions using multicontrast MRI improves efficiency and reproducibility versus manual delineation, and deep learning often reaches strong performance. Existing DL approaches, however, fail to jointly optimize in-domain accuracy and out-of-domain generalization when trained on a single limited source. This work introduces UNISELF, using test-time self-ensembled lesion fusion and test-time instance normalization of latent features to handle domain shifts and missing contrasts. Trained on ISBI 2015, it ranks among top methods and outperforms benchmarks across MICCAI 2016, UMCL, and a private multisite dataset.",{"@graph":69,"@context":122},[70,84,105],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":35,"@type":76,"position":81},"https://docshare.wps.com/document/healthcare/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/uniself-uniself-a-unified-network-with-instance-normalization-and-self-ensembled-lesion-fusion-for-multiple-sclerosis-lesion-segmentation/406904/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/uniself-uniself-a-unified-network-with-instance-normalization-and-self-ensembled-lesion-fusion-for-multiple-sclerosis-lesion-segmentation/406904.png","ImageObject",300,407,{"name":92,"@type":93},"pop out","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-11-03","2026-09-28",true,{"@type":102,"interactionType":103,"userInteractionCount":44},"InteractionCounter",{"@type":104},"ViewAction",{"@type":106,"mainEntity":107},"FAQPage",[108,114,118],{"name":109,"@type":110,"acceptedAnswer":111},"What problem does UNISELF address in MS lesion segmentation?","Question",{"text":112,"@type":113},"UNISELF targets the gap where deep learning methods do not simultaneously achieve strong in-domain accuracy and reliable out-of-domain generalization when trained on limited data from a single source.","Answer",{"name":115,"@type":110,"acceptedAnswer":116},"How does UNISELF improve robustness to domain shifts and missing input contrasts?",{"text":117,"@type":113},"UNISELF uses test-time self-ensembled lesion fusion to boost segmentation accuracy and test-time instance normalization of latent features to mitigate domain shifts and missing contrasts.",{"name":119,"@type":110,"acceptedAnswer":120},"How was UNISELF evaluated and what results were reported?",{"text":121,"@type":113},"UNISELF was trained on the ISBI 2015 longitudinal MS segmentation challenge training set and ranked among the best on the challenge test set, while outperforming benchmark methods on multiple out-of-domain datasets including MICCAI 2016, UMCL, and a private multisite dataset.","https://schema.org",{"og:url":83,"og:type":124,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":126,"canonical":83},"index,follow",{"doc_id":128,"site_id":62},406904,1791319258,{"code":4,"msg":5,"data":131},{"doc_id":128,"user_id":132,"nickname":92,"user_avatar":133,"doc_module":4,"category_id":34,"category_name":35,"doc_title":65,"doc_description":67,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":44,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":139,"language":140,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":67,"update_tm":144,"read_time":145,"preview_page_count":4},962090760408,"https://ap-avatar.wpscdn.com/davatar_085a072bc5b1113ac321206ff7593b45","arXiv :2508 .03982v1 [ ee ss .IV] 6 Aug 2025  \nUNISELF: A Unified Network with Instance Normalization and Self-Ensembled Lesion Fusion for Multiple Sclerosis Lesion Segmentation  \nJinwei Zhang 1a , Lianrui Zuob , Blake E. Deweyc , Samuel W. Remediosd , Yihao Liub , Savannah P. Haysa , Dzung L. Phame , Ellen M. Mowryc , Scott D. Newsomec , Peter A. Calabresic , Aaron Carassa , Jerry L. Princea  \na Department of Electrical and Computer Engineering, Johns Hopkins University, Baltimore, MD 21218, USA b Department of Electrical and Computer Engineering, Vanderbilt University, Nashville, TN 37215, USAc Department of Neurology, Johns Hopkins School of Medicine, Baltimore, MD 21287, USA  \nd Department of Computer Science, Johns Hopkins University, Baltimore, MD 21218, USA e Department of Radiology, Uniformed Services University of the Health Sciences, Bethesda, MD, 20814, USA  \nAbstract  \nAutomated segmentation of multiple sclerosis (MS) lesions using multicontrast magnetic resonance (MR) images improves efficiency and reproducibility compared to manual delineation, with deep learning (DL) methods achieving state-of-the-art performance. However, these DL-based methods have yet to simultaneously optimize in-domain accuracy and out-of-domain generalization when trained on a single source with limited data, or their performance has been unsatisfactory. To fill this gap, we propose a method called UNISELF, which achieves high accuracy within a single training domain while demonstrating strong generalizability across multiple out-of-domain test datasets. UNISELF employs a novel test-time self-ensembled lesion fusion to improve segmentation accuracy, and leverages test-time instance normalization (TTIN) of latent features to address domain shifts and missing input contrasts. Trained on the ISBI 2015 longitudinal MS segmentation challenge training dataset, UNISELF ranks among the best-performing methods on the challenge test dataset. Additionally, UNISELF outperforms all benchmark methods trained on the same ISBI training data across diverse out-of-domain test datasets with domain shifts and missing contrasts, including the public MICCAI 2016 and UMCL datasets, as well as a private multisite dataset. These test datasets exhibit domain shifts and/or missing contrasts caused by variations in acquisition protocols, scanner types, and imaging artifacts arising from imperfect acquisition. Our code is available at [https://github.com/uponacceptance](https://github.com/uponacceptance).  \nKeywords: Self-Ensemble, Instance Normalization, Domain Generalization, Multiple Sclerosis, Lesion Segmentation  \n1. Introduction  \nMultiple sclerosis (MS) is characterized by chronic inflammatory demyelination and neurodegeneration of the central nervous system (Haider et al., 2016) . Magnetic resonance imaging (MRI) is commonly used for diagnosing and monitoring MS in clinics due to its sensitivity to focal tissue inflammation (Filippi et al., 2019) . In MS, demyelinating white matter lesions typically appear hyperintense on T2-weighted (T2w) and T2-weighted Fluid-Attenuated Inversion Recovery (FLAIR) images, and hypointense on T1-weighted (T1w) images. Advancements in MRI, such  \nas higher spatial resolution, improved signal-to-noise ratio, and biomarkers specific to the pathology of MS disease (e.g. iron rim lesions (Dal-Bianco et al., 2017)) make analyzing the longitudinal evolution of the progression of the lesion more informative. These advancements improve the reliability of treatment evaluations (Kolb et al., 2022) . Accurate lesion mask delineations serve as a foundational step for quantifying lesion volumes and extracting advanced biomarkers relevant to lesion progression, such as iron rim lesions, which are linked to chronic active inflammation in MS (Kaunzner et al., 2019) . However,  \narXiv preprint August 7, 2025  \nmanual delineation of MS lesions faces challenges in efficiency and reproducibility due to their heterogeneous shape, size, and location ","cbCaia3fdHowzBdV","https://ap.wps.com/l/cbCaia3fdHowzBdV","pdf",29561666,33,"English","# Introduction\n## Multiple sclerosis, MRI, and the role of lesion segmentation\n## Challenges of manual delineation and dataset scarcity\n## Automated segmentation approaches: classical vs. learning-based\n## Deep learning progress and motivation for UNISELF","[{\"question\":\"What problem does UNISELF address in MS lesion segmentation?\",\"answer\":\"UNISELF targets the gap where deep learning methods do not simultaneously achieve strong in-domain accuracy and reliable out-of-domain generalization when trained on limited data from a single source.\"},{\"question\":\"How does UNISELF improve robustness to domain shifts and missing input contrasts?\",\"answer\":\"UNISELF uses test-time self-ensembled lesion fusion to boost segmentation accuracy and test-time instance normalization of latent features to mitigate domain shifts and missing contrasts.\"},{\"question\":\"How was UNISELF evaluated and what results were reported?\",\"answer\":\"UNISELF was trained on the ISBI 2015 longitudinal MS segmentation challenge training set and ranked among the best on the challenge test set, while outperforming benchmark methods on multiple out-of-domain datasets including MICCAI 2016, UMCL, and a private multisite dataset.\"}]","UNISELF - UNISELF - A Unified Network with Instance Normalization and Self-Ensembled Lesion Fusion for Multiple Sclerosis Lesion Segmentation | PDF",1790568862,83]