Rim Tyler Hyungtaek, Soh Zhi Da, Tham Yih-Chung, Yang Henrik Hee Seung, Lee Geunyoung, Kim Youngnam, Nusinovici Simon, Ting Daniel Shu Wei, Wong Tien Yin, Cheng Ching-Yu
Singapore Eye Research Institute, Singapore National Eye Centre, Singapore; Ophthalmology & Visual Sciences Academic Clinical Program (Eye ACP), Duke-NUS Medical School, Singapore.
Singapore Eye Research Institute, Singapore National Eye Centre, Singapore.
Ophthalmol Retina. 2020 Aug;4(8):793-800. doi: 10.1016/j.oret.2020.03.007. Epub 2020 Mar 17.
Though the domain of big data and artificial intelligence in health care continues to evolve, there is a lack of systemic methods to improve data quality and streamline the preparation process. To address this, we aimed to develop an automated sorting system (RetiSort) that accurately labels the type and laterality of retinal photographs.
Cross-sectional study.
RetiSort was developed with retinal photographs from the Singapore Epidemiology of Eye Diseases (SEED) study.
The development of RetiSort was composed of 3 steps: 2 deep-learning (DL) algorithms and 1 rule-based classifier. For step 1, a DL algorithm was developed to locate the optic disc, the "landmark feature." For step 2, based on the location of the optic disc derived from step 1, a rule-based classifier was developed to sort retinal photographs into 3 types: macular-centered, optic disc-centered, or related to other fields. Step 2 concurrently distinguished laterality (i.e., the left or right eye) of macular-centered photographs. For step 3, an additional DL algorithm was developed to differentiate the laterality of disc-centered photographs. Via the 3 steps, RetiSort sorted and labeled retinal images into (1) right macular-centered, (2) left macular-centered, (3) right optic disc-centered, (4) left optic disc-centered, and (5) images relating to other fields. Subsequently, the accuracy of RetiSort was evaluated on 5000 randomly selected retinal images from SEED as well as on 3 publicly available image databases (DIARETDB0, HEI-MED, and Drishti-GS). The main outcome measure was the accuracy for sorting of retinal photographs.
RetiSort mislabeled 48 out of 5000 retinal images from SEED, representing an overall accuracy of 99.0% (95% confidence interval [CI], 98.7-99.3). In external tests, RetiSort mislabeled 1, 0, and 2 images, respectively, from DIARETDB0, HEI-MED, and Drishti-GS, representing an accuracy of 99.2% (95% CI, 95.8-99.9), 100%, and 98.0% (95% CI, 93.1-99.8), respectively. Saliency maps consistently showed that the DL algorithm in step 3 required pixels in the central left lateral border and optic disc of optic disc-centered retinal photographs to differentiate the laterality.
RetiSort is a highly accurate automated sorting system. It can aid in data preparation and has practical applications in DL research that uses retinal photographs.
尽管大数据和人工智能在医疗保健领域不断发展，但仍缺乏提高数据质量和简化准备过程的系统方法。为解决这一问题，我们旨在开发一种自动分类系统（RetiSort），该系统能准确标记视网膜照片的类型和方位。
横断面研究。
RetiSort是利用新加坡眼病流行病学（SEED）研究中的视网膜照片开发的。
RetiSort的开发包括3个步骤：2种深度学习（DL）算法和1种基于规则的分类器。第一步，开发一种DL算法来定位视盘，即“地标特征”。第二步，基于第一步得出的视盘位置，开发一种基于规则的分类器，将视网膜照片分为3种类型：黄斑中心型、视盘中心型或与其他区域相关型。第二步同时区分黄斑中心型照片的方位（即左眼或右眼）。第三步，开发另一种DL算法来区分视盘中心型照片的方位。通过这3个步骤，RetiSort将视网膜图像分类并标记为：（1）右眼黄斑中心型，（2）左眼黄斑中心型，（3）右眼视盘中心型，（4）左眼视盘中心型，以及（5）与其他区域相关的图像。随后，在从SEED中随机选择的5000张视网膜图像以及3个公开可用的图像数据库（DIARETDB0、HEI-MED和Drishti-GS）上评估RetiSort的准确性。主要结局指标是视网膜照片分类的准确性。
RetiSort在来自SEED的5000张视网膜图像中错误标记了48张，总体准确率为99.0%（95%置信区间[CI]，98.7 - 99.3）。在外部测试中，RetiSort分别在DIARETDB0、HEI-MED和Drishti-GS中错误标记了1张、0张和2张图像，准确率分别为99.2%（95% CI，95.8 - 99.9）、100%和98.0%（95% CI，93.1 - 99.8）。显著性图一致显示，第三步中的DL算法需要视盘中心型视网膜照片的中央左侧边界和视盘中的像素来区分方位。
RetiSort是一种高度准确的自动分类系统。它有助于数据准备，并且在使用视网膜照片的DL研究中有实际应用。
