[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-82296-en":3,"doc-seo-82296-105":29,"detail-sidebar-cat-0-en-105":91},{"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":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},82296,13056703019662,"Evangeline","https://ap-avatar.wpscdn.com/avatar/be000253a8e92610077?_k=1778726343310543188",7,"Healthcare","From Classification to Localization and Clinical Validation Large-Scale Development of a Deep Learning System for Thoracic Disease Detection on Chest Radiographs in Thailand","Chest radiography (CXR) remains the most widely used thoracic imaging modality, but specialist reading is limited by a severe shortage of radiologists in Thailand and Southeast Asia. Local adaptation of deep learning models to Thai datasets can improve accuracy on local populations. The work presents development and comprehensive validation of Inspectra CXR v5, a model for multi-label thoracic disease classification and weakly supervised lesion localization. Trained on 874,858 radiographs, it achieved high AUROC and reliable cross-hospital generalization. Localization and radiologist usability evaluations show strong performance and concordance.","arXiv :2607 .09305v1 [ cs .CV] 10 Jul 2026  \nFROM CLASSIFICATION TO LOCALIZATION AND CLINICAL VALIDATION: LARGE-SCALE DEVELOPMENT OF A DEEP LEARNING SYSTEM FOR THORACIC DISEASE DETECTION ON CHEST RADIOGRAPHS IN THAILAND  \nA PREPRINT  \nIsarun Chamveha 1 , Tretap Promwiset 1 , Napat Wanchaitanawong 1 , Trongtum Tongdee3 , Pairash Saiviroonporn3 , and  \nWarasinee Chaisangmongkon2  \n1Perceptra Co., Ltd., Bangkok, Thailand  \n{isarun, tretap, [napat}@perceptra.tech](napat}@perceptra.tech)  \n2Institute of Field Robotics, King Mongkut’s University of Technology Thonburi, Bangkok, Thailand  \n[warasinee.cha@mail.kmutt.ac.th](warasinee.cha@mail.kmutt.ac.th)  \n3Radiology Department, Faculty of Medicine Siriraj Hospital, Mahidol University, Bangkok, Thailand  \n{trongtum, [pairash.sai}@gmail.com](pairash.sai}@gmail.com)  \nJuly 13, 2026  \nABSTRACT  \nChest radiography (CXR) remains the most widely used thoracic imaging modality, yet expert interpretation is constrained by a severe shortage of radiologists in Thailand and across Southeast Asia. Local adaptation of deep learning models to Thai data has been shown to substantially improve accuracy on Thai populations. Here we present the development and comprehensive validation of the chest radiograph analysis model in Inspectra CXR version 5, a deep learning system that performs multi-label thoracic disease classification and weakly supervised lesion localization within a single model. The architecture couples a DenseNet-121 backbone with Attend-and-Compare Modules (ACM) and a Probabilistic Class Activation Map (PCAM) aggregation layer, producing a per-condition classification score and heatmap simultaneously. The model was developed on  \n874,858 frontal chest radiographs with paired radiologist reports from Siriraj Hospital, Bangkok.  \nOn a held-out, radiologist-verified in-domain test set of 19,871 cases, it achieved a mean AUROC of 0.994 (mean sensitivity 92.4%, specificity 98.6%) across nine clinically important conditions.  \nOn an independent generalization set of 5,992 cases from 13 hospitals across Thailand, the mean AUROC was 0.970, indicating robust transfer across sites. For localization, evaluated on 4,549 radiologist-annotated cases, the model attained a mean lesion-localization fraction (LLF) of 77.9% at 0.59 non-lesion localizations per image. In a usability evaluation with five thoracic radiologists, the system reached a classification concordance of 93.6%, a localization concordance of 94.7%, anda mean System Usability Scale (SUS) score of 89 . These results indicate that a locally developed, localization-capable CXR system can deliver high accuracy, generalize across heterogeneous Thai hospitals, and earn the trust of practicing radiologists.  \nKeywords Deep Learning · Convolutional Neural Network · Chest X-Ray · Lesion Localization · Clinical Validation · Medical Image Analysis  \n1 Introduction  \nIn modern clinical practice, chest X-ray (CXR) is one of the most widely used methods for diagnosing abnormal conditions in the chest and nearby structures, owing to its noninvasive nature, low cost, and high availability. Chest  \nA PREPRINT-JULY 13, 2026  \nradiographs can reveal abnormalities in the lung parenchyma, mediastinum, rib cage, and heart, allowing physicians to determine the causes of various illnesses and monitor treatment for a range of life-threatening diseases such as pneumonia, tuberculosis, and cancer. In Western societies, CXR is performed on average 236 times per 1,000 patients, accounting for 25% of all diagnostic imaging procedures [1] .  \nIn Thailand and neighboring Southeast Asian countries, thoracic diseases constitute the leading causes of death across all age groups. Among the ten leading causes of death in Southeast Asia [2], five are thoracic abnormalities—tuberculosis, lung cancer, pneumonia, heart disease, and chronic obstructive pulmonary disease—all of which rely on chest X-ray imaging as a primary method of diagnosis. As the region has the highest rat","cbCaibKi8qhxswIO","https://ap.wps.com/l/cbCaibKi8qhxswIO","pdf",234281,1,12,"English","en",105,"# Abstract\n# Introduction\n## Clinical context and need for automated CXR analysis\n## Challenges in expert interpretation and generalizability\n## Motivation for local adaptation and deep learning","[{\"question\":\"Why is automated chest radiograph analysis needed in Thailand and Southeast Asia?\",\"answer\":\"Expert CXR interpretation is constrained by a shortage of radiologists, while thoracic diseases contribute heavily to mortality in the region. Scaling specialist reading is difficult and costly.\"},{\"question\":\"What does Inspectra CXR version 5 do in this system?\",\"answer\":\"It performs multi-label thoracic disease classification and weakly supervised lesion localization in a single model. The architecture couples a DenseNet-121 backbone with ACM and a PCAM aggregation layer to output classification scores and heatmaps.\"},{\"question\":\"How was the model validated and what performance was reported?\",\"answer\":\"It was trained on 874,858 frontal chest radiographs and evaluated on a radiologist-verified in-domain test set of 19,871 cases and an independent multi-hospital generalization set. Localization was assessed on radiologist-annotated cases, with additional usability evaluation involving thoracic radiologists.\"}]",1784179453,30,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":27},"from-classification-to-localization-and-clinical-validation-large-scale-development-of-a-deep-learning-system-for-thoracic-disease-detection-on-chest-radiographs-in-thailand","",{"@graph":35,"@context":85},[36,53,68],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/healthcare/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/from-classification-to-localization-and-clinical-validation-large-scale-development-of-a-deep-learning-system-for-thoracic-disease-detection-on-chest-radiographs-in-thailand/82296/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-17","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why is automated chest radiograph analysis needed in Thailand and Southeast Asia?","Question",{"text":75,"@type":76},"Expert CXR interpretation is constrained by a shortage of radiologists, while thoracic diseases contribute heavily to mortality in the region. Scaling specialist reading is difficult and costly.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What does Inspectra CXR version 5 do in this system?",{"text":80,"@type":76},"It performs multi-label thoracic disease classification and weakly supervised lesion localization in a single model. The architecture couples a DenseNet-121 backbone with ACM and a PCAM aggregation layer to output classification scores and heatmaps.",{"name":82,"@type":73,"acceptedAnswer":83},"How was the model validated and what performance was reported?",{"text":84,"@type":76},"It was trained on 874,858 frontal chest radiographs and evaluated on a radiologist-verified in-domain test set of 19,871 cases and an independent multi-hospital generalization set. Localization was assessed on radiologist-annotated cases, with additional usability evaluation involving thoracic radiologists.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,118,122,127,130,134],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":45,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":45,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":116,"slug":117},40,"healthcare",{"id":119,"doc_module":4,"doc_module_name":45,"category_name":120,"show_sort_weight":28,"slug":121},8,"Research & Report","research-report",{"id":123,"doc_module":4,"doc_module_name":45,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":45,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":45,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":45,"category_name":136,"show_sort_weight":106,"slug":137},19,"General","general"]