Automated Nonperfusion Quantification in Diabetic Retinopathy on Ultra-Widefield Swept-Source OCT Angiography

IF 5 Q1 OPHTHALMOLOGY

Ophthalmology science Pub Date : 2026-03-01 Epub Date: 2026-01-05 DOI:10.1016/j.xops.2025.101059

Tai Yong Loh MBBS , Juling Sia MBBS , Wei Hing Seah MBBS , Lingyi Zhuang MBBS , Wenjun Song MD , Yue Qiu , Xiaofeng Shen M.Eng , Zhongqing Yu M.Eng , Ryan Tan MBBS , Nuo Tang MBBS , Yusra Asad MBBS , Colin Ming Hui Goh MBBS , Charmayne Xinyi Ang MBBS , Celyn Chng MBBS , Peiqi Lo MBBS , Pavan Paniharam MBBS , Ser Koon Goh MBBS , Hnin Hnin Oo MBBS , Min Wang MD, PhD , Rupesh Agrawal MD , Sandy Wenting Zhou MD

{"title":"Automated Nonperfusion Quantification in Diabetic Retinopathy on Ultra-Widefield Swept-Source OCT Angiography","authors":"Tai Yong Loh MBBS , Juling Sia MBBS , Wei Hing Seah MBBS , Lingyi Zhuang MBBS , Wenjun Song MD , Yue Qiu , Xiaofeng Shen M.Eng , Zhongqing Yu M.Eng , Ryan Tan MBBS , Nuo Tang MBBS , Yusra Asad MBBS , Colin Ming Hui Goh MBBS , Charmayne Xinyi Ang MBBS , Celyn Chng MBBS , Peiqi Lo MBBS , Pavan Paniharam MBBS , Ser Koon Goh MBBS , Hnin Hnin Oo MBBS , Min Wang MD, PhD , Rupesh Agrawal MD , Sandy Wenting Zhou MD","doi":"10.1016/j.xops.2025.101059","DOIUrl":null,"url":null,"abstract":"<div><h3>Purpose</h3><div>To evaluate the performance of a customized deep learning algorithm for automated segmentation of nonperfusion area (NPA) on ultra-widefield swept-source OCTA (UWF SS-OCTA) and its utility in diabetic retinopathy (DR) severity assessment.</div></div><div><h3>Design</h3><div>Cross-sectional study.</div></div><div><h3>Subjects</h3><div>A total of 180 eyes from 122 participants representing all grades of DR severity.</div></div><div><h3>Methods</h3><div>We developed a convolutional neural network based on a multiscale U-Net backbone with squeeze-and-excitation attention for segmentation of NPAs on en face SS-OCTA all-retinal-layer images from 3 scan patterns: 6 × 6 mm, 12 × 12 mm, and 29 × 24 mm. Ground-truth annotations of NPAs and nongradable area (NGA) on en face OCTA images were generated by 2 independent graders and adjudicated by a vitreoretinal specialist. A corresponding en face structural OCT image was incorporated to distinguish true NPAs from shadow artifacts. Segmentation outputs included NPA, NGA, and shadow artifacts. Pixel-level accuracy was assessed with the F1 score. Nonperfusion index (NPI) was defined as NPA/gradable area. The level of agreement between human-labeled and algorithm-predicted NPI was analyzed using Bland–Altman analysis.</div></div><div><h3>Main Outcome Measures</h3><div>Algorithm F1 score and NPI.</div></div><div><h3>Results</h3><div>The algorithm for NPA segmentation achieved a mean F1 score of 0.82 ± 0.01 in 6 × 6 mm, 0.84 ± 0.03 in 12 × 12 mm, and 0.83 ± 0.02 in 29 × 24 mm, with no significant difference across fields of view (<em>P</em> = 0.12). Algorithm-derived NPI strongly agreed with expert grading (intraclass correlation coefficient >0.979). Both human- and algorithm-derived NPI increased progressively with increased DR severity in all scan patterns demonstrated by the Kruskal–Wallis test (6 × 6 mm: human: <em>P</em> = 0.02; algorithm: <em>P</em> = 0.03; 12 × 12 mm: algorithm <em>P</em> < 0.001; human <em>P</em> < 0.001; 29 × 24 mm: algorithm: <em>P</em> < 0.001; human: <em>P</em> < 0.001) with the largest magnitude of increase in 29 × 24 mm scans. The algorithm for foveal avascular zone segmentation also achieved a mean F1 score of 0.88 ± 0.05 for 6 × 6 mm images and 0.85 ± 0.05 for 12 × 12 mm images.</div></div><div><h3>Conclusions</h3><div>This deep learning algorithm was validated on single-scan UWF SS-OCTA for automated NPA segmentation and quantification. It demonstrates high accuracy and scalability across multiple scan sizes, supporting its potential integration into objective DR OCTA biomarker analysis.</div></div><div><h3>Financial Disclosure(s)</h3><div>Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.</div></div>","PeriodicalId":74363,"journal":{"name":"Ophthalmology science","volume":"6 3","pages":"Article 101059"},"PeriodicalIF":5.0000,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Ophthalmology science","FirstCategoryId":"1085","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2666914525003574","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2026/1/5 0:00:00","PubModel":"Epub","JCR":"Q1","JCRName":"OPHTHALMOLOGY","Score":null,"Total":0}

引用次数: 0

Abstract

Purpose

To evaluate the performance of a customized deep learning algorithm for automated segmentation of nonperfusion area (NPA) on ultra-widefield swept-source OCTA (UWF SS-OCTA) and its utility in diabetic retinopathy (DR) severity assessment.

Design

Cross-sectional study.

Subjects

A total of 180 eyes from 122 participants representing all grades of DR severity.

Methods

We developed a convolutional neural network based on a multiscale U-Net backbone with squeeze-and-excitation attention for segmentation of NPAs on en face SS-OCTA all-retinal-layer images from 3 scan patterns: 6 × 6 mm, 12 × 12 mm, and 29 × 24 mm. Ground-truth annotations of NPAs and nongradable area (NGA) on en face OCTA images were generated by 2 independent graders and adjudicated by a vitreoretinal specialist. A corresponding en face structural OCT image was incorporated to distinguish true NPAs from shadow artifacts. Segmentation outputs included NPA, NGA, and shadow artifacts. Pixel-level accuracy was assessed with the F1 score. Nonperfusion index (NPI) was defined as NPA/gradable area. The level of agreement between human-labeled and algorithm-predicted NPI was analyzed using Bland–Altman analysis.

Main Outcome Measures

Algorithm F1 score and NPI.

Results

The algorithm for NPA segmentation achieved a mean F1 score of 0.82 ± 0.01 in 6 × 6 mm, 0.84 ± 0.03 in 12 × 12 mm, and 0.83 ± 0.02 in 29 × 24 mm, with no significant difference across fields of view (P = 0.12). Algorithm-derived NPI strongly agreed with expert grading (intraclass correlation coefficient >0.979). Both human- and algorithm-derived NPI increased progressively with increased DR severity in all scan patterns demonstrated by the Kruskal–Wallis test (6 × 6 mm: human: P = 0.02; algorithm: P = 0.03; 12 × 12 mm: algorithm P < 0.001; human P < 0.001; 29 × 24 mm: algorithm: P < 0.001; human: P < 0.001) with the largest magnitude of increase in 29 × 24 mm scans. The algorithm for foveal avascular zone segmentation also achieved a mean F1 score of 0.88 ± 0.05 for 6 × 6 mm images and 0.85 ± 0.05 for 12 × 12 mm images.

Conclusions

This deep learning algorithm was validated on single-scan UWF SS-OCTA for automated NPA segmentation and quantification. It demonstrates high accuracy and scalability across multiple scan sizes, supporting its potential integration into objective DR OCTA biomarker analysis.

Financial Disclosure(s)

Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

查看原文本刊更多论文

超宽视场扫描源OCT血管造影在糖尿病视网膜病变中的自动非灌注定量

目的评价一种自定义深度学习算法在超宽视场扫描源OCTA （UWF - SS-OCTA）上非灌注区自动分割（NPA）的性能及其在糖尿病视网膜病变（DR）严重程度评估中的应用。DesignCross-sectional研究。受试者：来自122名参与者的180只眼睛，代表了所有程度的DR严重程度。方法基于多尺度U-Net骨干网，结合挤压-激发关注，构建了一种卷积神经网络，对6 × 6mm、12 × 12mm和29 × 24mm三种扫描模式的SS-OCTA全视网膜层图像进行npa分割。正面OCTA图像的npa和不可分级区域（NGA）的Ground-truth注释由2名独立分级员生成，并由一位玻璃体视网膜专家进行评审。结合相应的正面结构OCT图像来区分真正的npa和阴影伪影。分割输出包括NPA、NGA和阴影伪影。用F1评分评估像素级精度。非灌注指数（NPI）定义为NPA/可分级面积。使用Bland-Altman分析分析人工标记和算法预测NPI之间的一致程度。主要观察指标：算法F1评分和NPI。结果NPA分割算法在6 × 6 mm、12 × 12 mm和29 × 24 mm图像上的F1平均得分分别为0.82±0.01、0.84±0.03和0.83±0.02，视场间差异无统计学意义（P = 0.12）。算法导出的NPI与专家评分高度一致（类内相关系数>；0.979）。在Kruskal-Wallis测试显示的所有扫描模式中，人类和算法衍生的NPI都随着DR严重程度的增加而逐渐增加（6 × 6 mm：人类：P = 0.02；算法：P = 0.03； 12 × 12 mm：算法P <； 0.001；人类P <； 0.001； 29 × 24 mm：算法：P <； 0.001；人类：P <； 0.001），其中29 × 24 mm扫描增加幅度最大。该算法对6 × 6 mm图像的F1平均评分为0.88±0.05，对12 × 12 mm图像的F1平均评分为0.85±0.05。结论该深度学习算法在单扫描UWF - SS-OCTA上得到了验证，可用于NPA的自动分割和量化。它在多种扫描尺寸上具有很高的准确性和可扩展性，支持其与客观DR OCTA生物标志物分析的潜在集成。财务披露专有或商业披露可在本文末尾的脚注和披露中找到。

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