基于视觉显著性白内障检测的不同眼成像方式深度学习模型的多重比较

IF 4.6 Q1 OPHTHALMOLOGY

Ophthalmology science Pub Date : 2025-06-03 DOI:10.1016/j.xops.2025.100837

Jocelyn Hui Lin Goh BEng , Xiaofeng Lei MSc , Miao-Li Chee MPH , Yiming Qian PhD , Marco Yu PhD , Tyler Hyungtaek Rim MD, PhD , Simon Nusinovici PhD , David Ziyou Chen MBBS, FRCOphth , Kai Hui Koh BSc , Samantha Min Er Yew BSc , Yibing Chen BEng , Victor Teck Chang Koh MBBS, MMed , Charumathi Sabanayagam MD, PhD , Tien Yin Wong MD, PhD , Xinxing Xu PhD , Rick Siow Mong Goh PhD , Yong Liu PhD , Ching-Yu Cheng MD, PhD , Yih-Chung Tham PhD

{"title":"基于视觉显著性白内障检测的不同眼成像方式深度学习模型的多重比较","authors":"Jocelyn Hui Lin Goh BEng , Xiaofeng Lei MSc , Miao-Li Chee MPH , Yiming Qian PhD , Marco Yu PhD , Tyler Hyungtaek Rim MD, PhD , Simon Nusinovici PhD , David Ziyou Chen MBBS, FRCOphth , Kai Hui Koh BSc , Samantha Min Er Yew BSc , Yibing Chen BEng , Victor Teck Chang Koh MBBS, MMed , Charumathi Sabanayagam MD, PhD , Tien Yin Wong MD, PhD , Xinxing Xu PhD , Rick Siow Mong Goh PhD , Yong Liu PhD , Ching-Yu Cheng MD, PhD , Yih-Chung Tham PhD","doi":"10.1016/j.xops.2025.100837","DOIUrl":null,"url":null,"abstract":"<div><h3>Purpose</h3><div>Age-related cataract is the leading cause of vision impairment. Researchers have utilized various imaging modalities, including slit beam, diffuse anterior segment, and retinal imaging, to develop deep learning (DL) algorithms for automated cataract analysis. However, the comparative performance of these algorithms across different ocular imaging modalities remains unevaluated, mainly due to the absence of standardized test sets across studies.</div></div><div><h3>Design</h3><div>Retrospective study.</div></div><div><h3>Participants</h3><div>Across all the models, the Singapore Malay Eye Study data set was used for training (N = 7093 eyes) and internal testing (N = 1649 eyes). The Singapore Indian Eye Study (SINDI; N = 5579 eyes) and the Singapore Chinese Eye Study (SCES; N = 5658 eyes) were used for external testing. A community study data set of nonmydriatic retinal photos (N = 310 eyes) was used for external testing of the retinal model.</div></div><div><h3>Methods</h3><div>We developed 3 single-modality DL models (retinal, slit beam, and diffuse anterior segment photos) and 4 ensemble models (4 different combinations of the 3 single-modality models) to detect visually significant cataract (VSC). We defined eyes with VSC as having significant cataract (based on the modified Wisconsin cataract grading system) with a best-corrected visual acuity of <20/60.</div></div><div><h3>Main Outcome Measures</h3><div>Area under receiver operating characteristic curve (AUC).</div></div><div><h3>Results</h3><div>In the internal test, the retinal model had the highest AUC value (97.0%; 95% confidence interval [CI], 95.9–98.2), compared with the slit beam model (AUC, 93.4%; 95% CI, 90.1–96.7; <em>P</em><sub>diff</sub> = .029) and diffuse anterior segment model (AUC, 94.4; 95% CI, 92.3–96.4; <em>P</em><sub>diff</sub> = .002). There was no significant difference in AUC when comparing the retinal model with the ensemble models (all <em>P</em><sub>diff</sub> ≥ .07). These trends were consistently observed in the external test sets. In nonmydriatic eyes, the retinal model showed reasonable performance (AUC, 89.8%; 95% CI, 89.6–89.9).</div></div><div><h3>Conclusions</h3><div>Our findings highlight the retinal model as a promising tool for detecting VSC, outperforming slit beam and diffuse anterior segment models. Because retinal photography is routine in diabetic retinopathy screening, this approach could enable opportunistic cataract screening with minimal add-on cost.</div></div><div><h3>Financial Disclosure</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":"5 6","pages":"Article 100837"},"PeriodicalIF":4.6000,"publicationDate":"2025-06-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Multi-Comparison of Different Ocular Imaging Modality-based Deep Learning Models for Visually Significant Cataract Detection\",\"authors\":\"Jocelyn Hui Lin Goh BEng , Xiaofeng Lei MSc , Miao-Li Chee MPH , Yiming Qian PhD , Marco Yu PhD , Tyler Hyungtaek Rim MD, PhD , Simon Nusinovici PhD , David Ziyou Chen MBBS, FRCOphth , Kai Hui Koh BSc , Samantha Min Er Yew BSc , Yibing Chen BEng , Victor Teck Chang Koh MBBS, MMed , Charumathi Sabanayagam MD, PhD , Tien Yin Wong MD, PhD , Xinxing Xu PhD , Rick Siow Mong Goh PhD , Yong Liu PhD , Ching-Yu Cheng MD, PhD , Yih-Chung Tham PhD\",\"doi\":\"10.1016/j.xops.2025.100837\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><h3>Purpose</h3><div>Age-related cataract is the leading cause of vision impairment. Researchers have utilized various imaging modalities, including slit beam, diffuse anterior segment, and retinal imaging, to develop deep learning (DL) algorithms for automated cataract analysis. However, the comparative performance of these algorithms across different ocular imaging modalities remains unevaluated, mainly due to the absence of standardized test sets across studies.</div></div><div><h3>Design</h3><div>Retrospective study.</div></div><div><h3>Participants</h3><div>Across all the models, the Singapore Malay Eye Study data set was used for training (N = 7093 eyes) and internal testing (N = 1649 eyes). The Singapore Indian Eye Study (SINDI; N = 5579 eyes) and the Singapore Chinese Eye Study (SCES; N = 5658 eyes) were used for external testing. A community study data set of nonmydriatic retinal photos (N = 310 eyes) was used for external testing of the retinal model.</div></div><div><h3>Methods</h3><div>We developed 3 single-modality DL models (retinal, slit beam, and diffuse anterior segment photos) and 4 ensemble models (4 different combinations of the 3 single-modality models) to detect visually significant cataract (VSC). We defined eyes with VSC as having significant cataract (based on the modified Wisconsin cataract grading system) with a best-corrected visual acuity of <20/60.</div></div><div><h3>Main Outcome Measures</h3><div>Area under receiver operating characteristic curve (AUC).</div></div><div><h3>Results</h3><div>In the internal test, the retinal model had the highest AUC value (97.0%; 95% confidence interval [CI], 95.9–98.2), compared with the slit beam model (AUC, 93.4%; 95% CI, 90.1–96.7; <em>P</em><sub>diff</sub> = .029) and diffuse anterior segment model (AUC, 94.4; 95% CI, 92.3–96.4; <em>P</em><sub>diff</sub> = .002). There was no significant difference in AUC when comparing the retinal model with the ensemble models (all <em>P</em><sub>diff</sub> ≥ .07). These trends were consistently observed in the external test sets. In nonmydriatic eyes, the retinal model showed reasonable performance (AUC, 89.8%; 95% CI, 89.6–89.9).</div></div><div><h3>Conclusions</h3><div>Our findings highlight the retinal model as a promising tool for detecting VSC, outperforming slit beam and diffuse anterior segment models. Because retinal photography is routine in diabetic retinopathy screening, this approach could enable opportunistic cataract screening with minimal add-on cost.</div></div><div><h3>Financial Disclosure</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\":\"5 6\",\"pages\":\"Article 100837\"},\"PeriodicalIF\":4.6000,\"publicationDate\":\"2025-06-03\",\"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/S2666914525001356\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"OPHTHALMOLOGY\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Ophthalmology science","FirstCategoryId":"1085","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2666914525001356","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"OPHTHALMOLOGY","Score":null,"Total":0}

引用次数: 0

摘要

目的性白内障是视力受损的主要原因。研究人员利用各种成像模式，包括狭缝光束、弥漫性前段和视网膜成像，来开发用于自动白内障分析的深度学习（DL）算法。然而，这些算法在不同眼成像模式下的比较性能仍然没有得到评估，主要是由于缺乏标准化的测试集。DesignRetrospective研究。在所有模型中，新加坡马来眼研究数据集用于训练（N = 7093只眼睛）和内部测试（N = 1649只眼睛）。新加坡印度眼科研究(SINDI；N = 5579只眼睛)和新加坡华人眼研究(SCES；N = 5658只眼)进行外测。非散瞳视网膜照片的社区研究数据集（N = 310眼）用于视网膜模型的外部测试。方法建立3种单模态DL模型（视网膜、裂隙束和弥散前段照片）和4种集合模型（3种单模态模型的4种不同组合），用于检测显著性白内障（VSC）。我们将VSC定义为有明显白内障（基于改良的威斯康星白内障分级系统），最佳矫正视力为20/60。主要观察指标：受试者工作特征曲线下面积（AUC）。结果在内测中，视网膜模型的AUC值最高(97.0%；95%置信区间[CI]， 95.9-98.2)，与狭缝束模型相比(AUC, 93.4%；95% ci, 90.1-96.7；Pdiff = 0.029)和弥漫性前段模型(AUC, 94.4；95% ci, 92.3-96.4；Pdiff = .002)。视网膜模型与集合模型的AUC差异无统计学意义（均Pdiff≥0.07）。这些趋势在外部测试集中得到了一致的观察。在非散瞳眼，视网膜模型表现合理(AUC, 89.8%；95% ci, 89.6-89.9)。结论我们的研究结果表明，视网膜模型是一种很有前途的检测VSC的工具，优于狭缝束和弥漫性前段模型。由于视网膜摄影在糖尿病视网膜病变筛查中是常规的，这种方法可以以最小的额外成本进行机会性白内障筛查。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。

查看原文本刊更多论文

Multi-Comparison of Different Ocular Imaging Modality-based Deep Learning Models for Visually Significant Cataract Detection

Purpose

Age-related cataract is the leading cause of vision impairment. Researchers have utilized various imaging modalities, including slit beam, diffuse anterior segment, and retinal imaging, to develop deep learning (DL) algorithms for automated cataract analysis. However, the comparative performance of these algorithms across different ocular imaging modalities remains unevaluated, mainly due to the absence of standardized test sets across studies.

Design

Retrospective study.

Participants

Across all the models, the Singapore Malay Eye Study data set was used for training (N = 7093 eyes) and internal testing (N = 1649 eyes). The Singapore Indian Eye Study (SINDI; N = 5579 eyes) and the Singapore Chinese Eye Study (SCES; N = 5658 eyes) were used for external testing. A community study data set of nonmydriatic retinal photos (N = 310 eyes) was used for external testing of the retinal model.

Methods

We developed 3 single-modality DL models (retinal, slit beam, and diffuse anterior segment photos) and 4 ensemble models (4 different combinations of the 3 single-modality models) to detect visually significant cataract (VSC). We defined eyes with VSC as having significant cataract (based on the modified Wisconsin cataract grading system) with a best-corrected visual acuity of <20/60.

Main Outcome Measures

Area under receiver operating characteristic curve (AUC).

Results

In the internal test, the retinal model had the highest AUC value (97.0%; 95% confidence interval [CI], 95.9–98.2), compared with the slit beam model (AUC, 93.4%; 95% CI, 90.1–96.7; P_diff = .029) and diffuse anterior segment model (AUC, 94.4; 95% CI, 92.3–96.4; P_diff = .002). There was no significant difference in AUC when comparing the retinal model with the ensemble models (all P_diff ≥ .07). These trends were consistently observed in the external test sets. In nonmydriatic eyes, the retinal model showed reasonable performance (AUC, 89.8%; 95% CI, 89.6–89.9).

Conclusions

Our findings highlight the retinal model as a promising tool for detecting VSC, outperforming slit beam and diffuse anterior segment models. Because retinal photography is routine in diabetic retinopathy screening, this approach could enable opportunistic cataract screening with minimal add-on cost.