Yuting Wen , Yao Huang , Yu Liu , Shasha Zhang , Zhe Liu , Chan Hui , Yi Wang
{"title":"人工智能诊断幽门螺杆菌感染的内镜和病理组织图像：系统回顾和荟萃分析","authors":"Yuting Wen , Yao Huang , Yu Liu , Shasha Zhang , Zhe Liu , Chan Hui , Yi Wang","doi":"10.1016/j.ibmed.2025.100244","DOIUrl":null,"url":null,"abstract":"<div><h3>Background</h3><div>In recent years, artificial intelligence (AI) algorithms, including deep learning, have shown remarkable progress in image-recognition tasks. This study aimed to evaluate the diagnostic performance of AI in diagnosing Helicobacter pylori (H. pylori) infection using endoscopic and pathological images.</div></div><div><h3>Methods</h3><div>A literature search was conducted across multiple databases to identify all primary studies related to the diagnostic performance of AI algorithms for H. pylori infection published before 2024. True positive (TP), false positive (FP), false negative (FN), and true negative (TN) values were extracted or calculated for each study. Pooled sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), precision-recall (PR), and diagnostic odds ratio (DOR) were calculated. A summary receiver operating characteristic curve (SROC) was used to assess overall diagnostic performance.</div></div><div><h3>Results</h3><div>Twelve studies were included in the final analysis. The pooled sensitivity was 0.87 (95 % CI 0.78–0.92), pooled specificity was 0.79 (95 % CI 0.54–0.92), pooled PLR was 4.1 (95 % CI 1.7–9.8), and pooled NLR was 0.17 (95 % CI 0.10–0.29). The DOR was 24 (95 % CI 7–78), and the SROC was 0.90 (95 % CI 0.87–0.92). Substantial heterogeneity was observed among the studies (sensitivity: I<sup>2</sup> = 90.50 %, 95 % CI 86.37–94.62; specificity: I<sup>2</sup> = 98.66 %, 95 % CI 98.34–98.97). Deek's funnel plot indicated low publication bias (P = 0.89).</div></div><div><h3>Conclusions</h3><div>AI algorithms show potential in diagnosing HP infection by improving accuracy and lesion detection. However, due to heterogeneity in study results, more comprehensive clinical validation is needed before widespread application. Future research should focus on multicenter validation, standardized datasets, integration into clinical workflows, and addressing data privacy and ethics to promote broader use of AI in HP diagnosis.</div></div>","PeriodicalId":73399,"journal":{"name":"Intelligence-based medicine","volume":"11 ","pages":"Article 100244"},"PeriodicalIF":0.0000,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Artificial intelligence for the diagnosis of Helicobacter pylori infection in endoscopic and pathological tissues images: A systematic review and meta-analysis\",\"authors\":\"Yuting Wen , Yao Huang , Yu Liu , Shasha Zhang , Zhe Liu , Chan Hui , Yi Wang\",\"doi\":\"10.1016/j.ibmed.2025.100244\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><h3>Background</h3><div>In recent years, artificial intelligence (AI) algorithms, including deep learning, have shown remarkable progress in image-recognition tasks. This study aimed to evaluate the diagnostic performance of AI in diagnosing Helicobacter pylori (H. pylori) infection using endoscopic and pathological images.</div></div><div><h3>Methods</h3><div>A literature search was conducted across multiple databases to identify all primary studies related to the diagnostic performance of AI algorithms for H. pylori infection published before 2024. True positive (TP), false positive (FP), false negative (FN), and true negative (TN) values were extracted or calculated for each study. Pooled sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), precision-recall (PR), and diagnostic odds ratio (DOR) were calculated. A summary receiver operating characteristic curve (SROC) was used to assess overall diagnostic performance.</div></div><div><h3>Results</h3><div>Twelve studies were included in the final analysis. The pooled sensitivity was 0.87 (95 % CI 0.78–0.92), pooled specificity was 0.79 (95 % CI 0.54–0.92), pooled PLR was 4.1 (95 % CI 1.7–9.8), and pooled NLR was 0.17 (95 % CI 0.10–0.29). The DOR was 24 (95 % CI 7–78), and the SROC was 0.90 (95 % CI 0.87–0.92). Substantial heterogeneity was observed among the studies (sensitivity: I<sup>2</sup> = 90.50 %, 95 % CI 86.37–94.62; specificity: I<sup>2</sup> = 98.66 %, 95 % CI 98.34–98.97). Deek's funnel plot indicated low publication bias (P = 0.89).</div></div><div><h3>Conclusions</h3><div>AI algorithms show potential in diagnosing HP infection by improving accuracy and lesion detection. However, due to heterogeneity in study results, more comprehensive clinical validation is needed before widespread application. Future research should focus on multicenter validation, standardized datasets, integration into clinical workflows, and addressing data privacy and ethics to promote broader use of AI in HP diagnosis.</div></div>\",\"PeriodicalId\":73399,\"journal\":{\"name\":\"Intelligence-based medicine\",\"volume\":\"11 \",\"pages\":\"Article 100244\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2025-01-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Intelligence-based medicine\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S2666521225000481\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Intelligence-based medicine","FirstCategoryId":"1085","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2666521225000481","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
摘要
近年来，包括深度学习在内的人工智能（AI）算法在图像识别任务中取得了显著进展。本研究旨在评价人工智能对幽门螺杆菌（h.p ylori）感染的内镜和病理诊断效果。方法对多个数据库进行文献检索，筛选2024年之前发表的与人工智能算法诊断幽门螺杆菌感染性能相关的所有初步研究。提取或计算每项研究的真阳性（TP）、假阳性（FP）、假阴性（FN）和真阴性（TN）值。计算合并敏感性、特异性、阳性似然比（PLR）、阴性似然比（NLR）、查全率（PR）和诊断优势比（DOR）。综合受试者工作特征曲线（SROC）用于评估总体诊断表现。结果12项研究纳入最终分析。合并敏感性为0.87 (95% CI 0.78 ~ 0.92)，合并特异性为0.79 (95% CI 0.54 ~ 0.92)，合并PLR为4.1 (95% CI 1.7 ~ 9.8)，合并NLR为0.17 （95% CI 0.10 ~ 0.29）。DOR为24 (95% CI 7-78)， SROC为0.90 （95% CI 0.87-0.92）。研究间存在显著的异质性(敏感性：I2 = 90.50%, 95% CI 86.37-94.62；特异性：I2 = 98.66%, 95% CI 98.34-98.97)。Deek漏斗图显示低发表偏倚（P = 0.89）。结论ai算法在HP感染诊断中具有提高准确率和病灶检出率的潜力。但由于研究结果存在异质性，在广泛应用前需要更全面的临床验证。未来的研究应侧重于多中心验证、标准化数据集、整合到临床工作流程中，以及解决数据隐私和伦理问题，以促进人工智能在HP诊断中的广泛应用。
Artificial intelligence for the diagnosis of Helicobacter pylori infection in endoscopic and pathological tissues images: A systematic review and meta-analysis
Background
In recent years, artificial intelligence (AI) algorithms, including deep learning, have shown remarkable progress in image-recognition tasks. This study aimed to evaluate the diagnostic performance of AI in diagnosing Helicobacter pylori (H. pylori) infection using endoscopic and pathological images.
Methods
A literature search was conducted across multiple databases to identify all primary studies related to the diagnostic performance of AI algorithms for H. pylori infection published before 2024. True positive (TP), false positive (FP), false negative (FN), and true negative (TN) values were extracted or calculated for each study. Pooled sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), precision-recall (PR), and diagnostic odds ratio (DOR) were calculated. A summary receiver operating characteristic curve (SROC) was used to assess overall diagnostic performance.
Results
Twelve studies were included in the final analysis. The pooled sensitivity was 0.87 (95 % CI 0.78–0.92), pooled specificity was 0.79 (95 % CI 0.54–0.92), pooled PLR was 4.1 (95 % CI 1.7–9.8), and pooled NLR was 0.17 (95 % CI 0.10–0.29). The DOR was 24 (95 % CI 7–78), and the SROC was 0.90 (95 % CI 0.87–0.92). Substantial heterogeneity was observed among the studies (sensitivity: I2 = 90.50 %, 95 % CI 86.37–94.62; specificity: I2 = 98.66 %, 95 % CI 98.34–98.97). Deek's funnel plot indicated low publication bias (P = 0.89).
Conclusions
AI algorithms show potential in diagnosing HP infection by improving accuracy and lesion detection. However, due to heterogeneity in study results, more comprehensive clinical validation is needed before widespread application. Future research should focus on multicenter validation, standardized datasets, integration into clinical workflows, and addressing data privacy and ethics to promote broader use of AI in HP diagnosis.
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号
