Deep Learning in High-Resolution Anoscopy: Assessing the Impact of Staining and Therapeutic Manipulation on Automated Detection of Anal Cancer Precursors.

IF 3 3区医学 Q2 GASTROENTEROLOGY & HEPATOLOGY

Clinical and Translational Gastroenterology Pub Date : 2024-04-01 DOI:10.14309/ctg.0000000000000681

Miguel Mascarenhas Saraiva, Lucas Spindler, Nadia Fathallah, Hélene Beaussier, Célia Mamma, Tiago Ribeiro, João Afonso, Mariana Carvalho, Rita Moura, Pedro Cardoso, Francisco Mendes, Miguel Martins, Julien Adam, João Ferreira, Guilherme Macedo, Vincent de Parades

{"title":"Deep Learning in High-Resolution Anoscopy: Assessing the Impact of Staining and Therapeutic Manipulation on Automated Detection of Anal Cancer Precursors.","authors":"Miguel Mascarenhas Saraiva, Lucas Spindler, Nadia Fathallah, Hélene Beaussier, Célia Mamma, Tiago Ribeiro, João Afonso, Mariana Carvalho, Rita Moura, Pedro Cardoso, Francisco Mendes, Miguel Martins, Julien Adam, João Ferreira, Guilherme Macedo, Vincent de Parades","doi":"10.14309/ctg.0000000000000681","DOIUrl":null,"url":null,"abstract":"Introduction: High-resolution anoscopy (HRA) is the gold standard for detecting anal squamous cell carcinoma (ASCC) precursors. Preliminary studies on the application of artificial intelligence (AI) models to this modality have revealed promising results. However, the impact of staining techniques and anal manipulation on the effectiveness of these algorithms has not been evaluated. We aimed to develop a deep learning system for automatic differentiation of high-grade squamous intraepithelial lesion vs low-grade squamous intraepithelial lesion in HRA images in different subsets of patients (nonstained, acetic acid, lugol, and after manipulation).Methods: A convolutional neural network was developed to detect and differentiate high-grade and low-grade anal squamous intraepithelial lesions based on 27,770 images from 103 HRA examinations performed in 88 patients. Subanalyses were performed to evaluate the algorithm's performance in subsets of images without staining, acetic acid, lugol, and after manipulation of the anal canal. The sensitivity, specificity, accuracy, positive and negative predictive values, and area under the curve were calculated.Results: The convolutional neural network achieved an overall accuracy of 98.3%. The algorithm had a sensitivity and specificity of 97.4% and 99.2%, respectively. The accuracy of the algorithm for differentiating high-grade squamous intraepithelial lesion vs low-grade squamous intraepithelial lesion varied between 91.5% (postmanipulation) and 100% (lugol) for the categories at subanalysis. The area under the curve ranged between 0.95 and 1.00.Discussion: The introduction of AI to HRA may provide an accurate detection and differentiation of ASCC precursors. Our algorithm showed excellent performance at different staining settings. This is extremely important because real-time AI models during HRA examinations can help guide local treatment or detect relapsing disease.","PeriodicalId":10278,"journal":{"name":"Clinical and Translational Gastroenterology","volume":" ","pages":"e00681"},"PeriodicalIF":3.0000,"publicationDate":"2024-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11042766/pdf/","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Clinical and Translational Gastroenterology","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.14309/ctg.0000000000000681","RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"GASTROENTEROLOGY & HEPATOLOGY","Score":null,"Total":0}

引用次数: 0

Abstract

Introduction: High-resolution anoscopy (HRA) is the gold standard for detecting anal squamous cell carcinoma (ASCC) precursors. Preliminary studies on the application of artificial intelligence (AI) models to this modality have revealed promising results. However, the impact of staining techniques and anal manipulation on the effectiveness of these algorithms has not been evaluated. We aimed to develop a deep learning system for automatic differentiation of high-grade squamous intraepithelial lesion vs low-grade squamous intraepithelial lesion in HRA images in different subsets of patients (nonstained, acetic acid, lugol, and after manipulation).

Methods: A convolutional neural network was developed to detect and differentiate high-grade and low-grade anal squamous intraepithelial lesions based on 27,770 images from 103 HRA examinations performed in 88 patients. Subanalyses were performed to evaluate the algorithm's performance in subsets of images without staining, acetic acid, lugol, and after manipulation of the anal canal. The sensitivity, specificity, accuracy, positive and negative predictive values, and area under the curve were calculated.

Results: The convolutional neural network achieved an overall accuracy of 98.3%. The algorithm had a sensitivity and specificity of 97.4% and 99.2%, respectively. The accuracy of the algorithm for differentiating high-grade squamous intraepithelial lesion vs low-grade squamous intraepithelial lesion varied between 91.5% (postmanipulation) and 100% (lugol) for the categories at subanalysis. The area under the curve ranged between 0.95 and 1.00.

Discussion: The introduction of AI to HRA may provide an accurate detection and differentiation of ASCC precursors. Our algorithm showed excellent performance at different staining settings. This is extremely important because real-time AI models during HRA examinations can help guide local treatment or detect relapsing disease.

查看原文本刊更多论文

高分辨率肛门镜中的深度学习：评估染色和治疗操作对自动检测肛门癌前兆的影响。

简介：高分辨率肛门镜检查（HRA）是检测肛门鳞状细胞癌（ASCC）前兆的黄金标准。关于将人工智能（AI）模型应用于该模式的初步研究显示了良好的结果。然而，染色技术和肛门操作对这些算法有效性的影响尚未得到评估。我们旨在开发一种深度学习系统，用于自动区分不同亚组患者（未染色、醋酸、鲁戈尔和操作后）HRA 图像中的高级别（HSIL）与低级别（LSIL）鳞状上皮内病变：方法：根据 88 名患者 103 次 HRA 检查的 27,770 张图像，开发了一种卷积神经网络 (CNN)，用于检测和区分高级别和低级别肛门鳞状上皮内病变。我们还进行了子分析，以评估算法在无染色、醋酸、鲁戈尔和肛管操作后的图像子集中的性能。计算了灵敏度、特异性、准确性、阳性和阴性预测值以及曲线下面积（AUC）：结果：CNN 的总体准确率为 98.3%。该算法的灵敏度和特异度分别为 97.4% 和 99.2%。该算法区分 HSIL 和 LSIL 的准确率介于 91.5%（操作后）和 100%（鲁戈）之间。AUC介于0.95和1.00之间：将人工智能引入 HRA 可以准确检测和区分 ASCC 前体。我们的算法在不同的染色设置下都表现出卓越的性能。这一点极为重要，因为 HRA 检查中的实时人工智能模型有助于指导局部治疗或检测复发疾病。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Clinical and Translational Gastroenterology GASTROENTEROLOGY & HEPATOLOGY-

CiteScore

7.00

自引率

0.00%

发文量

114

审稿时长

16 weeks

期刊介绍： Clinical and Translational Gastroenterology (CTG), published on behalf of the American College of Gastroenterology (ACG), is a peer-reviewed open access online journal dedicated to innovative clinical work in the field of gastroenterology and hepatology. CTG hopes to fulfill an unmet need for clinicians and scientists by welcoming novel cohort studies, early-phase clinical trials, qualitative and quantitative epidemiologic research, hypothesis-generating research, studies of novel mechanisms and methodologies including public health interventions, and integration of approaches across organs and disciplines. CTG also welcomes hypothesis-generating small studies, methods papers, and translational research with clear applications to human physiology or disease. Colon and small bowel Endoscopy and novel diagnostics Esophagus Functional GI disorders Immunology of the GI tract Microbiology of the GI tract Inflammatory bowel disease Pancreas and biliary tract Liver Pathology Pediatrics Preventative medicine Nutrition/obesity Stomach.