AI-Enhanced OCT Analysis for Detecting ILM Removal in ERM Surgery.

IF 2.1 2区医学 Q2 OPHTHALMOLOGY

Retina-The Journal of Retinal and Vitreous Diseases Pub Date : 2025-07-08 DOI:10.1097/IAE.0000000000004606

Nehal Nailesh Mehta, An D Le, Ines D Nagel, Akshay Agnihotri, Anna Heinke, Lingyun Cheng, Dirk-Uwe Bartsch, Melanie Tran, Nguyen Truong, An Cheolhong, William R Freeman

{"title":"AI-Enhanced OCT Analysis for Detecting ILM Removal in ERM Surgery.","authors":"Nehal Nailesh Mehta, An D Le, Ines D Nagel, Akshay Agnihotri, Anna Heinke, Lingyun Cheng, Dirk-Uwe Bartsch, Melanie Tran, Nguyen Truong, An Cheolhong, William R Freeman","doi":"10.1097/IAE.0000000000004606","DOIUrl":null,"url":null,"abstract":"Purpose: This study evaluates how accurately humans and artificial intelligence (AI) can identify the type of surgery performed for epiretinal membrane (ERM) removal by analyzing postoperative optical coherence tomography (OCT) scans.Methods: A retrospective analysis at the University of California San Diego included 250 eyes from 239 patients who underwent vitrectomy for idiopathic ERM between January 2013 and October 2024. Eyes were categorized into two groups: one with both the internal limiting membrane (ILM) and ERM removed using indocyanine green (ICG) staining, and another with only ERM removal, guided by triamcinolone. Postoperative OCT scans were labeled as either ERM-only or ILM+ERM peel based on surgical notes. Both the human grader and AI model were trained on 200 labeled OCT scans and tested on 50 masked OCT scans to classify the surgery type.Results: Accuracy of the human grader in identifying the surgical technique was 50%, while the AI models demonstrated significantly higher accuracy. The ResNet18 model achieved 61±3%, while UwU-OrthLatt with DB4 initialization and UwU-PR-Relax with Symlet4 initialization reached 70±5% and 69±3%, respectively.Conclusions: AI outperformed human grading in detecting ILM removal from OCT scans, demonstrating AI's potential in improving ophthalmic imaging for clinical use.","PeriodicalId":54486,"journal":{"name":"Retina-The Journal of Retinal and Vitreous Diseases","volume":" ","pages":""},"PeriodicalIF":2.1000,"publicationDate":"2025-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Retina-The Journal of Retinal and Vitreous Diseases","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1097/IAE.0000000000004606","RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"OPHTHALMOLOGY","Score":null,"Total":0}

引用次数: 0

Abstract

Purpose: This study evaluates how accurately humans and artificial intelligence (AI) can identify the type of surgery performed for epiretinal membrane (ERM) removal by analyzing postoperative optical coherence tomography (OCT) scans.

Methods: A retrospective analysis at the University of California San Diego included 250 eyes from 239 patients who underwent vitrectomy for idiopathic ERM between January 2013 and October 2024. Eyes were categorized into two groups: one with both the internal limiting membrane (ILM) and ERM removed using indocyanine green (ICG) staining, and another with only ERM removal, guided by triamcinolone. Postoperative OCT scans were labeled as either ERM-only or ILM+ERM peel based on surgical notes. Both the human grader and AI model were trained on 200 labeled OCT scans and tested on 50 masked OCT scans to classify the surgery type.

Results: Accuracy of the human grader in identifying the surgical technique was 50%, while the AI models demonstrated significantly higher accuracy. The ResNet18 model achieved 61±3%, while UwU-OrthLatt with DB4 initialization and UwU-PR-Relax with Symlet4 initialization reached 70±5% and 69±3%, respectively.

Conclusions: AI outperformed human grading in detecting ILM removal from OCT scans, demonstrating AI's potential in improving ophthalmic imaging for clinical use.

查看原文本刊更多论文

人工智能增强的OCT分析在ERM手术中检测ILM去除。

目的：本研究通过分析术后光学相干断层扫描（OCT）扫描结果，评估人类和人工智能（AI）如何准确识别视网膜前膜（ERM）切除手术的类型。方法：加州大学圣地亚哥分校对2013年1月至2024年10月期间因特发性ERM接受玻璃体切除术的239例患者的250只眼睛进行回顾性分析。将眼睛分为两组：一组使用吲哚菁绿（ICG）染色去除内限制膜（ILM）和ERM，另一组在曲安奈德的指导下仅去除ERM。术后OCT扫描根据手术记录标记为仅ERM或ILM+ERM剥离。人类分级器和人工智能模型都在200个标记OCT扫描上进行了训练，并在50个屏蔽OCT扫描上进行了测试，以对手术类型进行分类。结果：人类评分员识别手术技术的准确率为50%，而人工智能模型的准确率明显更高。ResNet18模型达到61±3%，而初始化DB4的UwU-OrthLatt模型和初始化Symlet4的UwU-PR-Relax模型分别达到70±5%和69±3%。结论：人工智能在检测OCT扫描中的ILM去除方面优于人类分级，证明了人工智能在改善眼科成像临床应用方面的潜力。

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

Retina-The Journal of Retinal and Vitreous Diseases 医学-眼科学

CiteScore

5.70

自引率

9.10%

发文量

554

审稿时长

3-6 weeks

期刊介绍： RETINA® focuses exclusively on the growing specialty of vitreoretinal disorders. The Journal provides current information on diagnostic and therapeutic techniques. Its highly specialized and informative, peer-reviewed articles are easily applicable to clinical practice. In addition to regular reports from clinical and basic science investigators, RETINA® publishes special features including periodic review articles on pertinent topics, special articles dealing with surgical and other therapeutic techniques, and abstract cards. Issues are abundantly illustrated in vivid full color. Published 12 times per year, RETINA® is truly a “must have” publication for anyone connected to this field.