[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-86388-en":3,"doc-seo-86388-105":28,"detail-sidebar-cat-0-en-105":89},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":11,"language":21,"language_code":22,"site_id":23,"html_lang":22,"table_of_contents":24,"faqs":25,"seo_title":13,"seo_description":14,"update_tm":26,"read_time":27},86388,34359740700684,"Finn","https://ap-avatar.wpscdn.com/avatar/1f400023980c374ae676?_k=1777273430885731487",7,"Healthcare","Reliable Mislabel Detection for Video Capsule Endoscopy Data","Deep neural network classification depends on large, accurately annotated datasets, yet medical imaging makes reliable annotation difficult because specialized physicians must label data. This work introduces a mislabel-detection framework for medical datasets, addressing ambiguous class boundaries and inevitable label noise. Validation uses two large public Video Capsule Endoscopy datasets, Kvasir-Capsule and Galar. Potentially mislabeled samples are reviewed and re-annotated by three experienced gastroenterologists, improving anomaly detection after dataset cleaning.","Reliable Mislabel Detection for Video Capsule  \nEndoscopy Data  \n1st Julia Werner Embedded Systems University of Tu¨bingen T¨ubingen, Germany  \n5th Franz Brinkmann Department of Medicine I University Hospital Dresden TU Dresden, Germany  \n2nd Julius Oexle Embedded Systems University of Tu¨bingen T¨ubingen, Germany  \n6th Hannah Tolle Department of Medicine I University Hospital Dresden TU Dresden, Germany  \n3rd Oliver Bause Embedded Systems University of Tu¨bingen T¨ubingen, Germany  \n7th Jochen Hampe Department of Medicine I University Hospital Dresden TU Dresden, Germany  \n4th Maxime Le Floch Department of Medicine I University Hospital Dresden TU Dresden, Germany  \n8th Oliver Bringmann Embedded Systems University of Tu¨bingen T¨ubingen, Germany  \narXiv :2602 .06938v3 [ cs .CV] 13 Jul 2026  \nAbstract—The classification performance of deep neural networks relies strongly on access to large, accurately annotated datasets. In medical imaging, however, obtaining such datasets is particularly challenging since annotations must be provided by specialized physicians, which severely limits the pool of annotators. Furthermore, class boundaries can often be ambiguous or difficult to define which further complicates machine learningbased classification. In this paper, we want to address this problem and introduce a framework for mislabel detection in medical datasets. This is validated on the two largest, publicly available datasets for Video Capsule Endoscopy, an important imaging procedure for examining the gastrointestinal tract based on a video stream of low-resolution images. In addition, potentially mislabeled samples identified by our pipeline were reviewed and re-annotated by three experienced gastroenterologists. Our results show that the proposed framework successfully detects incorrectly labeled data and results in an improved anomaly detection performance after cleaning the datasets compared to current baselines.  \nIndex Terms—Video Capsule Endoscopy, Unsupervised Noise Detection, Anomaly Detection, Dataset Cleaning  \nI. INTRODUCTION  \nProficient performance of machine learning models highly depends on access to large, representative datasets. However, when large-scale datasets are annotated by humans, the occurence of mislabeled samples is inevitable and a general assumption that all annotations are accurate can introduce substantial issues. Since deep neural networks tend to overfit on noisy labels, such label noise can adversely affect the generalization and prediction performance [24], [28] . In medical applications, the annotation of real-world clinical data is particularly challenging, as it is time-consuming and typically requires specialized clinical expertise, which restricts the pool of qualified annotators.  \nAn example for such medical application is the Video Capsule Endoscopy (VCE), a key diagnostic medical procedure to examine the gastrointestinal (GI) tract, that was first introduced  \nThis work has been partly funded by the German Federal Ministry of Research, Technology and Space (BMFTR) in the project MEDGE (16ME0530) .  \nin the early 2000s [8], [27], [29] . The VCE is specifically applied to inspect the small intestine for pathologies while the stomach and colon can be investigated by standard procedures such as a gastroscopy or colonoscopy [3], [26] . This procedures involves a small pill-sized capsule consisting of a camera, a transmitter, a battery and LEDs, that can be swallowed by patients to record the inside of the digestive tract while it moves through the gastrointestinal organs (mouth, esophagus, stomach, small intestine, colon) [17], [18] . With current devices on the market, images recorded by the camera are directly transmitted to an on-body receiver for subsequent assessment by gastroenterologists [17] . For this application, the long-term objective is on-device anomaly detection in realtime to enable timely diagnosis.  \nTo realize successful screening for pathologies, vision models can be e","cbCaioz8HdKrBX4v","https://ap.wps.com/l/cbCaioz8HdKrBX4v","pdf",18379788,1,"English","en",105,"# Introduction\n# Related Work","[{\"question\":\"Why is mislabel detection important for Video Capsule Endoscopy datasets?\",\"answer\":\"Video Capsule Endoscopy datasets rely on physician annotations, which are time-consuming and limited by expert availability. Label noise and ambiguous class boundaries can degrade model generalization and anomaly detection performance, especially under anomaly class imbalance.\"},{\"question\":\"How is the proposed framework validated in the paper?\",\"answer\":\"The framework is evaluated on the two largest publicly available Video Capsule Endoscopy datasets: Kvasir-Capsule and Galar. A subset of flagged samples is reviewed and re-annotated by three experienced gastroenterologists.\"},{\"question\":\"What is the benefit of cleaning datasets using mislabel detection?\",\"answer\":\"Identifying incorrectly labeled samples and re-annotating them leads to improved anomaly detection performance. The results show better performance compared with baselines that use uncleaned datasets.\"}]",1784211446,18,{"code":4,"msg":29,"data":30},"ok",{"site_id":23,"language":22,"slug":31,"title":13,"keywords":32,"description":14,"schema_data":33,"social_meta":84,"head_meta":86,"extra_data":88,"updated_unix":26},"reliable-mislabel-detection-for-video-capsule-endoscopy-data","",{"@graph":34,"@context":83},[35,52,66],{"@type":36,"itemListElement":37},"BreadcrumbList",[38,42,46,49],{"item":39,"name":40,"@type":41,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":43,"name":44,"@type":41,"position":45},"https://docshare.wps.com/document/","Document",2,{"item":47,"name":12,"@type":41,"position":48},"https://docshare.wps.com/document/healthcare/",3,{"item":50,"name":13,"@type":41,"position":51},"https://docshare.wps.com/document/reliable-mislabel-detection-for-video-capsule-endoscopy-data/86388/",4,{"url":50,"name":13,"@type":53,"author":54,"headline":13,"publisher":56,"fileFormat":59,"inLanguage":22,"description":14,"dateModified":60,"datePublished":60,"encodingFormat":59,"isAccessibleForFree":61,"interactionStatistic":62},"DigitalDocument",{"name":9,"@type":55},"Person",{"url":39,"name":57,"@type":58},"DocShare","Organization","application/pdf","2026-07-16",true,{"@type":63,"interactionType":64,"userInteractionCount":4},"InteractionCounter",{"@type":65},"ViewAction",{"@type":67,"mainEntity":68},"FAQPage",[69,75,79],{"name":70,"@type":71,"acceptedAnswer":72},"Why is mislabel detection important for Video Capsule Endoscopy datasets?","Question",{"text":73,"@type":74},"Video Capsule Endoscopy datasets rely on physician annotations, which are time-consuming and limited by expert availability. Label noise and ambiguous class boundaries can degrade model generalization and anomaly detection performance, especially under anomaly class imbalance.","Answer",{"name":76,"@type":71,"acceptedAnswer":77},"How is the proposed framework validated in the paper?",{"text":78,"@type":74},"The framework is evaluated on the two largest publicly available Video Capsule Endoscopy datasets: Kvasir-Capsule and Galar. A subset of flagged samples is reviewed and re-annotated by three experienced gastroenterologists.",{"name":80,"@type":71,"acceptedAnswer":81},"What is the benefit of cleaning datasets using mislabel detection?",{"text":82,"@type":74},"Identifying incorrectly labeled samples and re-annotating them leads to improved anomaly detection performance. The results show better performance compared with baselines that use uncleaned datasets.","https://schema.org",{"og:url":50,"og:type":85,"og:title":13,"og:site_name":57,"og:description":14},"article",{"robots":87,"canonical":50},"index,follow",{"doc_id":7,"site_id":23},{"code":4,"msg":5,"data":90},[91,95,99,103,108,113,116,121,126,129,133],{"id":20,"doc_module":4,"doc_module_name":44,"category_name":92,"show_sort_weight":93,"slug":94},"Story & Novel",90,"story-novel",{"id":45,"doc_module":4,"doc_module_name":44,"category_name":96,"show_sort_weight":97,"slug":98},"Literature",80,"literature",{"id":51,"doc_module":4,"doc_module_name":44,"category_name":100,"show_sort_weight":101,"slug":102},"Exam",70,"exam",{"id":104,"doc_module":4,"doc_module_name":44,"category_name":105,"show_sort_weight":106,"slug":107},5,"Comic",60,"comic",{"id":109,"doc_module":4,"doc_module_name":44,"category_name":110,"show_sort_weight":111,"slug":112},6,"Technology",50,"technology",{"id":11,"doc_module":4,"doc_module_name":44,"category_name":12,"show_sort_weight":114,"slug":115},40,"healthcare",{"id":117,"doc_module":4,"doc_module_name":44,"category_name":118,"show_sort_weight":119,"slug":120},8,"Research & Report",30,"research-report",{"id":122,"doc_module":4,"doc_module_name":44,"category_name":123,"show_sort_weight":124,"slug":125},9,"Religion & Spirituality",20,"religion-spirituality",{"id":124,"doc_module":4,"doc_module_name":44,"category_name":127,"show_sort_weight":124,"slug":128},"World Cup","world-cup",{"id":130,"doc_module":4,"doc_module_name":44,"category_name":131,"show_sort_weight":130,"slug":132},10,"Lifestyle","lifestyle",{"id":134,"doc_module":4,"doc_module_name":44,"category_name":135,"show_sort_weight":104,"slug":136},19,"General","general"]