[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-370928-105":3,"detail-sidebar-cat-0-en-105":81,"doc-detail-370928-en":130},{"code":4,"msg":5,"data":6},0,"ok",{"site_id":7,"language":8,"slug":9,"title":10,"keywords":11,"description":12,"schema_data":13,"social_meta":74,"head_meta":76,"extra_data":78,"updated_unix":80},105,"en","development-of-deep-learning-based-narrow-band-imaging-endocytoscopic-classification-for-predicting-colorectal-lesions-from-a-retrospective-study","Development of deep learning-based narrow-band imaging endocytoscopic classification for predicting colorectal lesions from a retrospective study","","Data-driven methods have improved colorectal lesion diagnosis in digestive endoscopy, yet applications in endocytoscopy (EC)—a high-magnification imaging technique—remain limited, with many studies using conventional machine learning models. A computer-aided diagnosis (CAD) approach was developed using narrow-band imaging endocytoscopy to classify lesions into non-neoplastic lesions, adenomas, and invasive cancers. Using multi-stage pre-training with supervised deep clustering in a multicenter retrospective cohort, the model outperforms state-of-the-art supervised methods, improves diagnostic accuracy, and supports endoscopists as an assistive tool, enabling better consistency and early cancer screening.",{"@graph":14,"@context":73},[15,34,56],{"@type":16,"itemListElement":17},"BreadcrumbList",[18,23,27,31],{"item":19,"name":20,"@type":21,"position":22},"https://docshare.wps.com","Home","ListItem",1,{"item":24,"name":25,"@type":21,"position":26},"https://docshare.wps.com/document/","Document",2,{"item":28,"name":29,"@type":21,"position":30},"https://docshare.wps.com/document/research-report/","Research & Report",3,{"item":32,"name":10,"@type":21,"position":33},"https://docshare.wps.com/document/development-of-deep-learning-based-narrow-band-imaging-endocytoscopic-classification-for-predicting-colorectal-lesions-from-a-retrospective-study/370928/",4,{"url":32,"name":10,"@type":35,"image":36,"author":41,"headline":10,"publisher":44,"fileFormat":47,"inLanguage":8,"description":12,"dateModified":48,"datePublished":49,"encodingFormat":47,"isAccessibleForFree":50,"interactionStatistic":51},"DigitalDocument",{"url":37,"@type":38,"width":39,"height":40},"https://docshare.wps.com/thumbnails/development-of-deep-learning-based-narrow-band-imaging-endocytoscopic-classification-for-predicting-colorectal-lesions-from-a-retrospective-study/370928.png","ImageObject",300,407,{"name":42,"@type":43},"Ben ","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-10-24","2026-09-23",true,{"@type":52,"interactionType":53,"userInteractionCount":55},"InteractionCounter",{"@type":54},"ViewAction",9,{"@type":57,"mainEntity":58},"FAQPage",[59,65,69],{"name":60,"@type":61,"acceptedAnswer":62},"What problem does the study address in endocytoscopy-based colorectal diagnosis?","Question",{"text":63,"@type":64},"Existing data-driven approaches for colorectal lesions have limited application in endocytoscopy, where many prior works rely on conventional machine learning rather than stronger deep learning strategies.","Answer",{"name":66,"@type":61,"acceptedAnswer":67},"How does the proposed CAD model classify colorectal lesions?",{"text":68,"@type":64},"It classifies lesions into three categories: non-neoplastic lesions, adenomas, and invasive cancers, rather than only performing binary neoplastic versus non-neoplastic separation.",{"name":70,"@type":61,"acceptedAnswer":71},"What training strategy does the CAD model use and what is its reported performance?",{"text":72,"@type":64},"The model is trained with a multi-stage pre-training strategy combined with supervised deep clustering, and it outperforms state-of-the-art supervised methods in a multicenter retrospective cohort while surpassing endoscopists’ accuracy in human-machine competitions.","https://schema.org",{"og:url":32,"og:type":75,"og:title":10,"og:site_name":45,"og:description":12},"article",{"robots":77,"canonical":32},"index,follow",{"doc_id":79,"site_id":7},370928,1790889658,{"code":4,"msg":82,"data":83},"success",[84,88,92,96,101,106,111,115,119,122,126],{"id":22,"doc_module":4,"doc_module_name":25,"category_name":85,"show_sort_weight":86,"slug":87},"Story & Novel",90,"story-novel",{"id":26,"doc_module":4,"doc_module_name":25,"category_name":89,"show_sort_weight":90,"slug":91},"Literature",80,"literature",{"id":33,"doc_module":4,"doc_module_name":25,"category_name":93,"show_sort_weight":94,"slug":95},"Exam",70,"exam",{"id":97,"doc_module":4,"doc_module_name":25,"category_name":98,"show_sort_weight":99,"slug":100},5,"Comic",60,"comic",{"id":102,"doc_module":4,"doc_module_name":25,"category_name":103,"show_sort_weight":104,"slug":105},6,"Technology",50,"technology",{"id":107,"doc_module":4,"doc_module_name":25,"category_name":108,"show_sort_weight":109,"slug":110},7,"Healthcare",40,"healthcare",{"id":112,"doc_module":4,"doc_module_name":25,"category_name":29,"show_sort_weight":113,"slug":114},8,30,"research-report",{"id":55,"doc_module":4,"doc_module_name":25,"category_name":116,"show_sort_weight":117,"slug":118},"Religion & Spirituality",20,"religion-spirituality",{"id":117,"doc_module":4,"doc_module_name":25,"category_name":120,"show_sort_weight":117,"slug":121},"World Cup","world-cup",{"id":123,"doc_module":4,"doc_module_name":25,"category_name":124,"show_sort_weight":123,"slug":125},10,"Lifestyle","lifestyle",{"id":127,"doc_module":4,"doc_module_name":25,"category_name":128,"show_sort_weight":97,"slug":129},19,"General","general",{"code":4,"msg":82,"data":131},{"doc_id":79,"user_id":132,"nickname":42,"user_avatar":133,"doc_module":4,"category_id":112,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":55,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":139,"language":140,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":12,"update_tm":144,"read_time":113,"preview_page_count":33,"preview_urls":145},2336478951081,"https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c","Article [https://doi.org/10.1038/s41467-025-63812-5](https://doi.org/10.1038/s41467-025-63812-5)  \nDevelopment ofdeep learning-based narrowband imaging endocytoscopic classiﬁcation for predicting colorectal lesions from a retrospective study  \nReceived: 27 September 2024  \n\n| Accepted: 27 August 2025 |\n| --- |\n| |\n| Check for updates |\n\nJie Wang1,7, Mingqing Liu 1,7, Haiming Liao2, Jiawei Fan1, He Zhu1, Tantan Ma1, Dong Yang1, Fengming Ni1, Fan Zhang1, Guohua Jin1, Juan Yu3, Jiahui He4,5, Xiaokun Liang 5, Nan Zhang 1 , Hong Xu 1  & Zhicheng Zhang 1,5,6   \nData-driven approaches have advanced colorectal lesion diagnosis in digestive endoscopy, yet their application in endocytoscopy (EC)—a high-magniﬁcation imaging technique—remains limited, with most studies relying on conventional machine learning methods like support vector machines. Inspired by the success of large-scale language models that leverage progressive pre-training, we develop a computer-aided diagnosis (CAD) model using narrow-band imaging endocytoscopy (EC-NBI) to classify colorectal lesions (non-neoplastic lesions, adenomas, and invasive cancers). Here, we show that our model, trained through a multi-stage pre-training strategy combined with supervised deep clustering, outperforms state-of-the-art supervised methods in a multicenter retrospective cohort. Notably, it surpasses endoscopists’ diagnostic accuracy in human-machine competitions and enhances their performance when used as an assistive tool. This EC-NBI CAD model signiﬁcantly improves the accuracy and consistency of diagnosing colorectal lesions, laying a foundation for future early cancer screening, particularly for distinguishing superﬁcial and deep submucosal invasive cancers, pending further expansive multi-center data.  \nEarly disease screening is crucial for improving patient prognosis and reducing healthcare costs1,2, which can signiﬁcantly decrease morbidity and mortality rates by detecting diseases at an early, more treatable stage3,4. This is particularly crucial in the realm of colorectal lesions, including precancerous lesions and colorectal cancer (CRC). In this context, early detection via screening methods makes timely intervention and more favorable prognoses4–7. Endoscopy, as a noninvasive and highly accurate diagnostic tool, allows for direct visualization of the gastrointestinal tract8. It is instrumental in identifying  \nprecancerous lesions and early-stage cancers, facilitating timely treatment. Endocytoscopy (EC) represents an advanced endoscopic system that enables the observation of cellular atypia and glandular structures with ultra-high magniﬁcation (e.g., ×520) images following appropriate staining9,10. Additionally, EC combined with narrow-band imaging (EC-NBI) enables visualization of microvessels on the surface of colorectal lesions without the need for staining, thereby avoiding potential health risks associated with the staining process. Studies have demonstrated that EC-NBI provides diagnostic efﬁcacy in  \n1Department of Gastroenterology, The First Hospital of Jilin University, Changchun, China. 2Boluo County Health Bureau, Huizhou, China. 3Department of Radiology, The First Afﬁliated Hospital of Shenzhen University, Shenzhen, China. 4School of Computer Science, University of Nottingham Ningbo China, Ningbo, China. 5Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China. 6JancsiLab, JancsiTech, Hongkong, China. 7These authors contributed equally: Jie Wang, Mingqing Liu. e-mail: zhangnan@j[lu.edu.cn](lu.edu.cn); x_hong@j[lu.edu.cn](lu.edu.cn); [zhangzhicheng13@mails.ucas.edu.cn](zhangzhicheng13@mails.ucas.edu.cn)  \npredicting the histopathological diagnosis of colorectal lesions comparable to EC with staining11. Compared to non-magniﬁed NBI or magnifying NBI (ME-NBI, with magniﬁcation levels of approximately ×80–100), the ultra-high magniﬁcation capability of EC-NBI enables more precise observation of surface vascular patterns, facilitati","cbCaii1wpLK1i6hK","https://ap.wps.com/l/cbCaii1wpLK1i6hK","pdf",3188779,12,"English","# Background and Rationale\n## Need for EC-NBI beyond conventional machine learning\n## Importance of early colorectal disease screening\n# Proposed CAD Method\n## Three-class EC-NBI lesion classification\n## Multi-stage pre-training and supervised deep clustering\n# Results and Potential Impact\n## Performance versus state-of-the-art methods and human experts\n## Benefits as an assistive diagnostic tool\n## Future directions for multicenter data","[{\"question\":\"What problem does the study address in endocytoscopy-based colorectal diagnosis?\",\"answer\":\"Existing data-driven approaches for colorectal lesions have limited application in endocytoscopy, where many prior works rely on conventional machine learning rather than stronger deep learning strategies.\"},{\"question\":\"How does the proposed CAD model classify colorectal lesions?\",\"answer\":\"It classifies lesions into three categories: non-neoplastic lesions, adenomas, and invasive cancers, rather than only performing binary neoplastic versus non-neoplastic separation.\"},{\"question\":\"What training strategy does the CAD model use and what is its reported performance?\",\"answer\":\"The model is trained with a multi-stage pre-training strategy combined with supervised deep clustering, and it outperforms state-of-the-art supervised methods in a multicenter retrospective cohort while surpassing endoscopists’ accuracy in human-machine competitions.\"}]","Development of deep learning-based narrow-band imaging endocytoscopic classification for predicting colorectal lesions from a retrospective study | PDF",1790190847,[146,150,154,158],{"page_no":22,"url":147,"width":148,"height":149},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/72520c773cb7957ecb513c3c378e09c8.webp",983,1305,{"page_no":26,"url":151,"width":152,"height":153},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/6262aaf6f61bcaa4117d60f775c3a747.webp",982,1304,{"page_no":30,"url":155,"width":156,"height":157},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/7baa5b81b619ea054216baa44ad20fa1.webp",981,1303,{"page_no":33,"url":159,"width":160,"height":161},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/d7c272d77cbef8ce082f7108768efd05.webp",980,1301]