Patrícia Andrade, Miguel Mascarenhas, Francisco Mendes, Bruno Rosa, Pedro Cardoso, João Afonso, Tiago Ribeiro, Miguel Martins, Joana Mota, Maria João Almeida, Tiago Cúrdia Gonçalves, Pedro Campelo, Cláudia Macedo, António Pinto da Costa, Cecílio Santander, Jack di Palma, João Ferreira, José Cotter, Guilherme Macedo
{"title":"人工智能辅助胶囊内窥镜检测克罗恩病溃疡和糜烂：一项多中心验证研究","authors":"Patrícia Andrade, Miguel Mascarenhas, Francisco Mendes, Bruno Rosa, Pedro Cardoso, João Afonso, Tiago Ribeiro, Miguel Martins, Joana Mota, Maria João Almeida, Tiago Cúrdia Gonçalves, Pedro Campelo, Cláudia Macedo, António Pinto da Costa, Cecílio Santander, Jack di Palma, João Ferreira, José Cotter, Guilherme Macedo","doi":"10.1016/j.cgh.2025.09.036","DOIUrl":null,"url":null,"abstract":"<p><strong>Background and aims: </strong>Small bowel capsule endoscopy (SBCE) is limited by lengthy, variable interpretation. Artificial intelligence (AI) offers a transformative approach, enabling faster and more accurate lesion detection. This multicenter study aimed to validate an AI model for ulcers and erosions across different SBCE devices.</p><p><strong>Methods: </strong>A multicenter, cross-sectional cohort study was conducted from 2021 to 2024, involving centers in Europe and the USA. Two SBCE devices (PillCamSB3™ and Olympus EC-10®) were used. The performance of AI-assisted reading, generated by a deep learning model, was compared with standard-of-care (SoC) reading using a reference standard defined by an independent review board. The study utilized two SBCE devices (PillCamSB3™, Olympus EC-10®) and analyzed 259 SBCE exams. The performance of AI-assisted reading generated by the deep learning model was compared with standard of care (SoC) reading against a reference standard defined by an independent review board.</p><p><strong>Results: </strong>Ulcers and erosions were detected in 93 patients (35.9%). SoC had 69.6% sensitivity, 99.4% specificity, 98.5% PPV, 85.6% NPV, and 88.8% accuracy. AI-assisted reading detected ulcers and erosions with 90.2% sensitivity, 84.4% specificity, 76.1% PPV, 94.0% NPV, and 86.5% accuracy. The detection yield of AI-assisted reading was superior (p<0.001) to conventional SoC reading. The AI-assisted physician SBCE reading identified 568 lesions out of 600 identified by expert board review (94.7%). The median AI-assisted CE reporting time was 172 seconds per exam.</p><p><strong>Conclusions: </strong>The AI-assisted SBCE reading achieved superior diagnostic performance compared to SoC, with a substantial decrease in reading time.</p>","PeriodicalId":10347,"journal":{"name":"Clinical Gastroenterology and Hepatology","volume":" ","pages":""},"PeriodicalIF":12.0000,"publicationDate":"2025-10-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"AI-ASSISTED CAPSULE ENDOSCOPY FOR DETECTION OF ULCERS AND EROSIONS IN CROHN'S DISEASE: A MULTICENTER VALIDATION STUDY.\",\"authors\":\"Patrícia Andrade, Miguel Mascarenhas, Francisco Mendes, Bruno Rosa, Pedro Cardoso, João Afonso, Tiago Ribeiro, Miguel Martins, Joana Mota, Maria João Almeida, Tiago Cúrdia Gonçalves, Pedro Campelo, Cláudia Macedo, António Pinto da Costa, Cecílio Santander, Jack di Palma, João Ferreira, José Cotter, Guilherme Macedo\",\"doi\":\"10.1016/j.cgh.2025.09.036\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><strong>Background and aims: </strong>Small bowel capsule endoscopy (SBCE) is limited by lengthy, variable interpretation. Artificial intelligence (AI) offers a transformative approach, enabling faster and more accurate lesion detection. This multicenter study aimed to validate an AI model for ulcers and erosions across different SBCE devices.</p><p><strong>Methods: </strong>A multicenter, cross-sectional cohort study was conducted from 2021 to 2024, involving centers in Europe and the USA. Two SBCE devices (PillCamSB3™ and Olympus EC-10®) were used. The performance of AI-assisted reading, generated by a deep learning model, was compared with standard-of-care (SoC) reading using a reference standard defined by an independent review board. The study utilized two SBCE devices (PillCamSB3™, Olympus EC-10®) and analyzed 259 SBCE exams. The performance of AI-assisted reading generated by the deep learning model was compared with standard of care (SoC) reading against a reference standard defined by an independent review board.</p><p><strong>Results: </strong>Ulcers and erosions were detected in 93 patients (35.9%). SoC had 69.6% sensitivity, 99.4% specificity, 98.5% PPV, 85.6% NPV, and 88.8% accuracy. AI-assisted reading detected ulcers and erosions with 90.2% sensitivity, 84.4% specificity, 76.1% PPV, 94.0% NPV, and 86.5% accuracy. The detection yield of AI-assisted reading was superior (p<0.001) to conventional SoC reading. The AI-assisted physician SBCE reading identified 568 lesions out of 600 identified by expert board review (94.7%). The median AI-assisted CE reporting time was 172 seconds per exam.</p><p><strong>Conclusions: </strong>The AI-assisted SBCE reading achieved superior diagnostic performance compared to SoC, with a substantial decrease in reading time.</p>\",\"PeriodicalId\":10347,\"journal\":{\"name\":\"Clinical Gastroenterology and Hepatology\",\"volume\":\" \",\"pages\":\"\"},\"PeriodicalIF\":12.0000,\"publicationDate\":\"2025-10-09\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Clinical Gastroenterology and Hepatology\",\"FirstCategoryId\":\"3\",\"ListUrlMain\":\"https://doi.org/10.1016/j.cgh.2025.09.036\",\"RegionNum\":1,\"RegionCategory\":\"医学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"GASTROENTEROLOGY & HEPATOLOGY\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Clinical Gastroenterology and Hepatology","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1016/j.cgh.2025.09.036","RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"GASTROENTEROLOGY & HEPATOLOGY","Score":null,"Total":0}
AI-ASSISTED CAPSULE ENDOSCOPY FOR DETECTION OF ULCERS AND EROSIONS IN CROHN'S DISEASE: A MULTICENTER VALIDATION STUDY.
Background and aims: Small bowel capsule endoscopy (SBCE) is limited by lengthy, variable interpretation. Artificial intelligence (AI) offers a transformative approach, enabling faster and more accurate lesion detection. This multicenter study aimed to validate an AI model for ulcers and erosions across different SBCE devices.
Methods: A multicenter, cross-sectional cohort study was conducted from 2021 to 2024, involving centers in Europe and the USA. Two SBCE devices (PillCamSB3™ and Olympus EC-10®) were used. The performance of AI-assisted reading, generated by a deep learning model, was compared with standard-of-care (SoC) reading using a reference standard defined by an independent review board. The study utilized two SBCE devices (PillCamSB3™, Olympus EC-10®) and analyzed 259 SBCE exams. The performance of AI-assisted reading generated by the deep learning model was compared with standard of care (SoC) reading against a reference standard defined by an independent review board.
Results: Ulcers and erosions were detected in 93 patients (35.9%). SoC had 69.6% sensitivity, 99.4% specificity, 98.5% PPV, 85.6% NPV, and 88.8% accuracy. AI-assisted reading detected ulcers and erosions with 90.2% sensitivity, 84.4% specificity, 76.1% PPV, 94.0% NPV, and 86.5% accuracy. The detection yield of AI-assisted reading was superior (p<0.001) to conventional SoC reading. The AI-assisted physician SBCE reading identified 568 lesions out of 600 identified by expert board review (94.7%). The median AI-assisted CE reporting time was 172 seconds per exam.
Conclusions: The AI-assisted SBCE reading achieved superior diagnostic performance compared to SoC, with a substantial decrease in reading time.
期刊介绍：
Clinical Gastroenterology and Hepatology (CGH) is dedicated to offering readers a comprehensive exploration of themes in clinical gastroenterology and hepatology. Encompassing diagnostic, endoscopic, interventional, and therapeutic advances, the journal covers areas such as cancer, inflammatory diseases, functional gastrointestinal disorders, nutrition, absorption, and secretion.
As a peer-reviewed publication, CGH features original articles and scholarly reviews, ensuring immediate relevance to the practice of gastroenterology and hepatology. Beyond peer-reviewed content, the journal includes invited key reviews and articles on endoscopy/practice-based technology, health-care policy, and practice management. Multimedia elements, including images, video abstracts, and podcasts, enhance the reader's experience. CGH remains actively engaged with its audience through updates and commentary shared via platforms such as Facebook and Twitter.
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号
