[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-364922-105":59,"doc-detail-364922-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","investigation-on-potential-bias-factors-in-histopathology-datasets","Investigation on potential bias factors in histopathology datasets","","Deep neural networks show strong performance in digital pathology, but concerns persist that dataset biases may drive over-optimistic results rather than true histological signals. This study analyzes site-specific bias effects in features used for cancer-type discrimination, where models also achieve high accuracy for acquisition site classification. Two DNNs, KimiaNet and EfficientNet, are evaluated using balanced accuracy for data-center detection to identify potential causes of the unexpected site pattern recognition ability.",{"@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/investigation-on-potential-bias-factors-in-histopathology-datasets/364922/",{"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/investigation-on-potential-bias-factors-in-histopathology-datasets/364922.png","ImageObject",300,407,{"name":92,"@type":93},"Genevieve","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-11-17","2026-09-23",true,{"@type":102,"interactionType":103,"userInteractionCount":52},"InteractionCounter",{"@type":104},"ViewAction",{"@type":106,"mainEntity":107},"FAQPage",[108,114,118],{"name":109,"@type":110,"acceptedAnswer":111},"Why do deep learning models in digital pathology sometimes produce over-optimistic results?","Question",{"text":112,"@type":113},"They may rely on biased or dataset-specific cues rather than histological characteristics, leading to high in-distribution accuracy but weaker generalization.","Answer",{"name":115,"@type":110,"acceptedAnswer":116},"What does “site-specific bias” refer to in this research?",{"text":117,"@type":113},"The embedded features extracted for cancer-type discrimination can contain patterns tied to acquisition site, enabling high acquisition site classification accuracy.",{"name":119,"@type":110,"acceptedAnswer":120},"How are the models evaluated in the study?",{"text":121,"@type":113},"KimiaNet and a self-trained EfficientNet are assessed using balanced accuracy for classifying data centers, aiming to investigate factors behind site-specific pattern recognition.","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},364922,1791356238,{"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":52,"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":19,"preview_urls":146},1374391974585,"https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c","[www. nature.com/scientificreports](www. nature.com/scientificreports)  \nOPEN  \nInvestigation on potential bias factors in histopathology datasets  \nFarnaz Kheiri1􀀍, Shahryar Rahnamayan2, Masoud Makrehchi1 & Azam Asilian Bidgoli3  \nDeep neural networks (DNNs) have demonstrated remarkable capabilities in medical applications, including digital pathology, where they excel at analyzing complex patterns in medical images to assist in accurate disease diagnosis and prognosis. However, concerns have arisen about potential biases in The Cancer Genome Atlas (TCGA) dataset, a comprehensive repository of digitized histopathology data and serves as both a training and validation source for deep learning models, suggesting that over-optimistic results of model performance may be due to reliance on biased features rather than histological characteristics. Surprisingly, recent studies have confirmed the existence of site-specific bias in the embedded features extracted for cancer-type discrimination, leading to high accuracy in acquisition site classification. This biased behavior motivated us to conduct an in-depth analysis to investigate potential causes behind this unexpected biased ability toward site-specific pattern recognition. The analysis was conducted on two cutting-edge DNN models: KimiaNet, a state-of-theart DNN trained on TCGA images, and the self-trained EfficientNet. In this research study, the balanced accuracy metric is used to evaluate the performance of a model trained to classify data centers, which was originally designed to learn cancerous patterns, with the aim of investigating the potential factors contributing to the higher balanced accuracy in data center detection.  \nKeywords Digital pathology, Site-specific bias, Learning models, Histopathology  \nIn the rapidly developing landscape of Artificial Intelligence (AI) and Machine Learning (ML), we are becoming increasingly conscious of an emerging concern-the imperative effects of bias in these systems. As AI technologies revolutionize our lives and play a crucial role in making critical decisions, such as healthcare, recognizing the roots of biased behavior in AI models becomes essential for increasing the trustworthiness and generality of these models1–4. Emerging bias in deep models’s decision-making process refers to the phenomenon where an AI model is able to attain high accuracy on a particular task without actually learning the features required to deal with that task. In other words, the model relies on heuristics specific to the training data but does not generalize well to unseen data. This can be a significant challenge in real-world machine learning applications5–7. In the realm of histopathology, distinct categories of biases that arise from various sources8,9, as summarized in Fig. 1, can negatively affect definitive diagnosis and prognosis steps. Though specific terminologies have been provided to introduce them, these types of biases are closely related to the concept of ‘shortcuts’ in the broader machine learning literature. These biases can lead to satisfactory in-distribution performance but poor out-ofdistribution performance. The first category,“Sampling bias,” occurs when the target population is not randomly picked for the data obtained during model training. In other words, when training data applied to deep models are not impartially representative of the population intended to serve, sampling bias of data kind may appear. Then, the model may not perform equally well for all demographic groups due to underrepresented or overrepresented behavior of data in the training set10–12. Roach et al., for the first time, focused on AI prognostic models for digital pathology images from prostate cancer trials and assessed their equity in the context of racial disparities between African American (AA) and non-AA populations, with the dominant portion of the data relating to thenon-AA population. Their findings suggest the need for further diverse evaluati","cbCaidGcTKIuX3kZ","https://ap.wps.com/l/cbCaidGcTKIuX3kZ","pdf",2691357,18,"English","# Introduction\n## Bias in AI and ML decision-making\n## Bias categories and their impact in histopathology\n## Sampling bias and demographic disparities\n## Dataset bias and shortcut learning in medical imaging","[{\"question\":\"Why do deep learning models in digital pathology sometimes produce over-optimistic results?\",\"answer\":\"They may rely on biased or dataset-specific cues rather than histological characteristics, leading to high in-distribution accuracy but weaker generalization.\"},{\"question\":\"What does “site-specific bias” refer to in this research?\",\"answer\":\"The embedded features extracted for cancer-type discrimination can contain patterns tied to acquisition site, enabling high acquisition site classification accuracy.\"},{\"question\":\"How are the models evaluated in the study?\",\"answer\":\"KimiaNet and a self-trained EfficientNet are assessed using balanced accuracy for classifying data centers, aiming to investigate factors behind site-specific pattern recognition.\"}]","Investigation on potential bias factors in histopathology datasets | PDF",1790159440,45,[147,151,155,159],{"page_no":8,"url":148,"width":149,"height":150},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/a7c554e39b20b3b75e731dc95a6c375a.webp",983,1291,{"page_no":14,"url":152,"width":153,"height":154},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/5e30c2ec5044e0f5edad6b914e3f1e9d.webp",982,1290,{"page_no":81,"url":156,"width":157,"height":158},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/6adf60c98bc52bab0d4f58fd9ea73e76.webp",981,1289,{"page_no":19,"url":160,"width":161,"height":162},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/d2029a671fc0051458e1265e8cc0cf3d.webp",980,1287]