[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-271056-105":59,"doc-detail-271056-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","is-dataset-quality-still-a-concern-in-diagnosis-using-large-foundation-model","Is Dataset Quality Still a Concern in Diagnosis Using Large Foundation Model?","","Recent advancements in pre-trained large foundation models (LFM) have driven major progress across natural language processing and computer vision, with growing impact in medical diagnostic workflows. For fundus imaging, an LFM built with a Vision Transformer and self-supervised learning shows strong results across multiple datasets, yet dataset quality—especially image quality and dataset bias—remains a key risk in clinical data. This study evaluates whether LFM is more robust to quality variations and bias, and whether fine-tuning can reduce these effects, comparing performance to typical convolutional networks.",{"@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":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/is-dataset-quality-still-a-concern-in-diagnosis-using-large-foundation-model/271056/",{"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/is-dataset-quality-still-a-concern-in-diagnosis-using-large-foundation-model/271056.png","ImageObject",300,407,{"name":92,"@type":93},"McGucket","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-17","2026-09-15",true,{"@type":102,"interactionType":103,"userInteractionCount":39},"InteractionCounter",{"@type":104},"ViewAction",{"@type":106,"mainEntity":107},"FAQPage",[108,114,118],{"name":109,"@type":110,"acceptedAnswer":111},"What problem does the study address regarding large foundation models in diagnosis?","Question",{"text":112,"@type":113},"The study examines whether dataset quality—image quality and dataset bias—still matters when using large foundation models for medical diagnosis, specifically fundus disease diagnosis.","Answer",{"name":115,"@type":110,"acceptedAnswer":116},"How does the study evaluate image quality effects and dataset bias effects?",{"text":117,"@type":113},"It conducts experiments on two fundus diagnosis tasks using datasets of varying quality, validating performance on different quality subsets and different categories within the same datasets.",{"name":119,"@type":110,"acceptedAnswer":120},"Which fine-tuning strategy is effective for mitigating dataset quality issues?",{"text":121,"@type":113},"The study finds that overall fine-tuning effectively mitigates the impact of dataset quality issues, and it compares strategies including linear probe, updating the whole model, and Tip-Adapter.","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},271056,1791392946,{"code":4,"msg":5,"data":131},{"doc_id":128,"user_id":132,"nickname":92,"user_avatar":133,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":39,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":52,"language":139,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":140,"faqs":141,"seo_title":142,"seo_description":67,"update_tm":143,"read_time":144,"preview_page_count":4},1236954412713,"https://us-avatar.wpscdn.com/davatar_29158cc5080c5b710cf443261637dec0","arXiv :2405 . 12584v1 [ ee ss .IV] 21 May 2024  \nIs Dataset Quality Still a Concern in Diagnosis Using Large Foundation Model?  \nZiqin Lin 1 ,2∗, Heng Li 1∗B, Zinan Li 1 ,2 , Huazhu Fu3 , and Jiang Liu 1 ,4 ,5 B  \n1 Research Institute of Trustworthy Autonomous Systems, Southern University of Science and Technology, Shenzhen, China  \n2 Department of Computer Science and Engineering, Southern University of Science and Technology, Shenzhen, China  \n3 Institute of High Performance Computing, Agency for Science, Technology and Research, Singapore  \n4 Guangdong Provincial Key Laboratory of Brain-inspired Intelligent Computation, Southern University of Science and Technology, Shenzhen, China  \n5 Research Institute of Trustworthy Autonomous Systems, Southern University of Science and Technology, Shenzhen, China  \n6 Department of Electrical and Computer Engineering  \nlih3, [liuj@sustech.edu.cn](liuj@sustech.edu.cn)  \nAbstract. Recent advancements in pre-trained large foundation models (LFM) have yielded significant breakthroughs across various domains, including natural language processing and computer vision. These models have been particularly impactful in the domain of medical diagnostic tasks. With abundant unlabeled data, an LFM has been developed for fundus images using the Vision Transformer (VIT) and a self-supervised learning framework. This LFM has shown promising performance in fundus disease diagnosis across multiple datasets. On the other hand, deep learning models have long been challenged by dataset quality issues, such as image quality and dataset bias. To investigate the influence of data quality on LFM, we conducted explorations in two fundus diagnosis tasks using datasets of varying quality. Specifically, we explored the following questions: Is LFM more robust to image quality? Is LFM affected by dataset bias? Can fine-tuning techniques alleviate these effects? Our investigation found that LFM exhibits greater resilience to dataset quality issues, including image quality and dataset bias, compared to typical convolutional networks. Furthermore, we discovered that overall fine-tuning is an effective adapter for LFM to mitigate the impact of dataset quality issues.  \nKeywords: LFM, dataset quality, dataset bias, image quality.  \n1 Introduction  \nRecent advancements in LFM have revolutionized the field of natural language processing (NLP) and computer vision (CV) as a whole. These models, such as GPT-3 [1] and its successors, have achieved remarkable performance in various NLP tasks, including language generation, translation, summarization, and  \nquestion-answering. In the CV region, Meta introduced the Segment Anything Model (SAM) [7] as a versatile and scalable solution for image segmentation. Moreover, LFM has shown remarkable potential in revolutionizing healthcare by analyzing medical text, assisting clinical decision-making, accelerating drug discovery, and contributing to public health surveillance [2,5,6,11,15] . Therefore, due to the vast amount of fundus image data, Zhou et al. proposed RETFound [17], a foundational model for retinal images, which learns generalized representations from mass unlabeled retinal images and serves as a basis for efficient model adaptation to various applications with a few labeled data. Specifically, RETFound utilizes a modified version of VIT [3] as the base model, trained on 1.6 million unlabeled retinal images through self-supervised learning, followed by adaptation to disease detection tasks with a few explicit labels. RETFound consistently outperforms multiple comparative models in the diagnosis and prognosis of vision-threatening eye diseases, as well as event prediction for complex systemic diseases such as heart failure and myocardial infarction, despite having fewer labeled data.  \nOn the other hand, dataset quality remains a common issue in clinical practice [12,9,10], which can adversely affect both physicians’ observations and algorithmic diagnoses. The dataset qualit","cbCaiawSeuEszCEU","https://ap.wps.com/l/cbCaiawSeuEszCEU","pdf",1172172,"English","# Introduction\n## Contribution summary","[{\"question\":\"What problem does the study address regarding large foundation models in diagnosis?\",\"answer\":\"The study examines whether dataset quality—image quality and dataset bias—still matters when using large foundation models for medical diagnosis, specifically fundus disease diagnosis.\"},{\"question\":\"How does the study evaluate image quality effects and dataset bias effects?\",\"answer\":\"It conducts experiments on two fundus diagnosis tasks using datasets of varying quality, validating performance on different quality subsets and different categories within the same datasets.\"},{\"question\":\"Which fine-tuning strategy is effective for mitigating dataset quality issues?\",\"answer\":\"The study finds that overall fine-tuning effectively mitigates the impact of dataset quality issues, and it compares strategies including linear probe, updating the whole model, and Tip-Adapter.\"}]","Is Dataset Quality Still a Concern in Diagnosis Using Large Foundation Model? | PDF",1789459215,25]