Csizek Zsófia, Mikó-Baráth Eszter, Budai Anna, Frigyik Andrew B, Pusztai Ágota, Nemes Vanda A, Závori László, Fülöp Diána, Czigler András, Szabó-Guth Kitti, Buzás Péter, Piñero David P, Jandó Gábor
Institute of Physiology, Medical School, University of Pécs, Pécs, Hungary.
Centre for Neuroscience, University of Pécs, Pécs, Hungary.
Front Med (Lausanne). 2023 Dec 22;10:1294559. doi: 10.3389/fmed.2023.1294559. eCollection 2023.
The development of costs-effective and sensitive screening solutions to prevent amblyopia and identify its risk factors (strabismus, refractive problems or mixed) is a significant priority of pediatric ophthalmology. The main objective of our study was to compare the classification performance of various vision screening tests, including classic, stereoacuity-based tests (Lang II, TNO, Stereo Fly, and Frisby), and non-stereoacuity-based, low-density static, dynamic, and noisy anaglyphic random dot stereograms. We determined whether the combination of non-stereoacuity-based tests integrated in the simplest artificial intelligence (AI) model could be an alternative method for vision screening.
Our study, conducted in Spain and Hungary, is a non-experimental, cross-sectional diagnostic test assessment focused on pediatric eye conditions. Using convenience sampling, we enrolled 423 children aged 3.6-14 years, diagnosed with amblyopia, strabismus, or refractive errors, and compared them to age-matched emmetropic controls. Comprehensive pediatric ophthalmologic examinations ascertained diagnoses. Participants used filter glasses for stereovision tests and red-green goggles for an AI-based test over their prescribed glasses. Sensitivity, specificity, and the area under the ROC curve (AUC) were our metrics, with sensitivity being the primary endpoint. AUCs were analyzed using DeLong's method, and binary classifications (pathologic vs. normal) were evaluated using McNemar's matched pair and Fisher's nonparametric tests.
Four non-overlapping groups were studied: (1) amblyopia ( = 46), (2) amblyogenic ( = 55), (3) non-amblyogenic ( = 128), and (4) emmetropic ( = 194), and a fifth group that was a combination of the amblyopia and amblyogenic groups. Based on AUCs, the AI combination of non-stereoacuity-based tests showed significantly better performance 0.908, 95% CI: (0.829-0.958) for detecting amblyopia and its risk factors than most classical tests: Lang II: 0.704, (0.648-0.755), Stereo Fly: 0.780, (0.714-0.837), Frisby: 0.754 (0.688-0.812), < 0.02, = 91, DeLong's method). At the optimum ROC point, McNemar's test indicated significantly higher sensitivity in accord with AUCs. Moreover, the AI solution had significantly higher sensitivity than TNO ( = 0.046, N = 134, Fisher's test), as well, while the specificity did not differ.
The combination of multiple tests utilizing anaglyphic random dot stereograms with varying parameters (density, noise, dynamism) in AI leads to the most advanced and sensitive screening test for identifying amblyopia and amblyogenic conditions compared to all the other tests studied.
开发具有成本效益且灵敏的筛查方案以预防弱视并识别其危险因素（斜视、屈光问题或混合型）是小儿眼科的一项重要优先事项。我们研究的主要目的是比较各种视力筛查测试的分类性能，包括基于传统立体视锐度的测试（朗 II 型、TNO、立体蝇和弗里斯比）以及基于非立体视锐度的低密度静态、动态和噪声互补色随机点立体图。我们确定了整合在最简单人工智能（AI）模型中的基于非立体视锐度测试的组合是否可以作为视力筛查的替代方法。
我们在西班牙和匈牙利进行的研究是一项针对小儿眼部疾病的非实验性横断面诊断测试评估。采用便利抽样，我们招募了 423 名年龄在 3.6 - 14 岁之间、被诊断患有弱视、斜视或屈光不正的儿童，并将他们与年龄匹配的正视对照组进行比较。全面的小儿眼科检查确定诊断。参与者在其处方眼镜上佩戴滤光眼镜进行立体视觉测试，并佩戴红绿色护目镜进行基于 AI 的测试。敏感性、特异性和 ROC 曲线下面积（AUC）是我们的指标，敏感性是主要终点。使用德龙方法分析 AUC，使用麦克内马尔配对检验和费舍尔非参数检验评估二元分类（病理性与正常）。
研究了四个不重叠的组：（1）弱视组（n = 46），（2）致弱视组（n = 55），（3）非致弱视组（n = 128），和（4）正视组（n = 194），以及第五组，即弱视组和致弱视组的组合。基于 AUC，基于非立体视锐度测试的 AI 组合在检测弱视及其危险因素方面表现出明显优于大多数传统测试的性能：0.908，95%CI：（0.829 - 0.958），而朗 II 型为：0.704，（0.648 - 0.755），立体蝇为：0.780，（0.714 - 0.837），弗里斯比为：0.754（0.688 - 0.812），P < 0.02，df = 91，德龙方法）。在最佳 ROC 点，麦克内马尔检验表明与 AUC 一致的敏感性显著更高。此外，AI 解决方案的敏感性也显著高于 TNO（P = 0.046，N = 134，费舍尔检验），而特异性没有差异。
与所有其他研究的测试相比，在人工智能中结合使用具有不同参数（密度、噪声、动态性）的互补色随机点立体图的多项测试，可得出用于识别弱视和致弱视情况的最先进且灵敏的筛查测试。
