王萱 陈向东 刘志敏 欧阳红波

〔摘要〕 人工智能（artificial intelligence， AI）不断普及，为医学等传统行业的发展带来了前所未有的契机。眼科是一门以影像学诊断为主的学科，有着与AI深度结合的潜力。中医眼科的特色是整体观念和辨证论治，故将整体症状与舌脉象融入眼科AI系统，对疾病的诊断十分有必要。以部分眼病（糖尿病视网膜病变、早产儿视网膜病变）为例，从眼底病变的筛查与预测的重要性、远程医疗现状、眼底图像的采集及自动诊断、AI与中医辨证论治的结合等方面，介绍目前AI在部分眼底病变筛查中的新技术、新应用，阐述目前发展的不足与面临的挑战。

〔关键词〕 人工智能;眼底病变;中医眼科;糖尿病视网膜病变;早产儿视网膜病变

〔中图分类号〕R276.7       〔文献标志码〕A        〔文章编号〕doi：10.3969/j.issn.1674-070X.2021.12.028

Progress of Artificial Intelligence in Fundus Disease Screening

WANG Xuan1，2，3，4， CHEN Xiangdong1，3，4*， LIU Zhimin1，3，4， OUYANG Hongbo5

（1. The First Affiliated Hospital of Hunan University of Chinese Medicine， Changsha， Hunan 410007， China; 2. Hunan University of Chinese Medicine， Changsha， Hunan 410208， China; 3. Hunan Provincial Key Laboratory for Prevention and Treatment of Ophthalmology and Otolaryngology Diseases with Chinese Medicine， Changsha， Hunan 410208， China; 4. Hunan Provincial Engineering and Technological Research Center for Prevention and Treatment of Ophthalmology and Otolaryngology Diseases with Chinese Medicine and Protecting Visual Function， Changsha， Hunan 410208， China; 5. Anxiang County Third People’s Hospital， Changde， Hunan 415600， China）

〔Abstract〕 The continuous popularization of artificial intelligence （AI） has brought unprecedented opportunities for the

development of traditional industries such as medicine. Ophthalmology is a subject based on imaging diagnosis， which has the potential of deep integration with AI. The characteristics of ophthalmology of traditional Chinese medicine are holistic concept and syndrome differentiation and treatment. Therefore， it is necessary to integrate the overall symptoms and tongue pulse into the ophthalmic AI system for the diagnosis of diseases. Taking some eye diseases as an example， this paper introduces the new technology and application of AI in screening fundus lesions from the aspects of the importance of screening and prediction of fundus lesions， the status quo of telemedicine， the collection and automatic diagnosis of fundus images， and the combination of AI and TCM syndrome differentiation and treatment， and expounds the shortcomings and challenges of current development.

〔Keywords〕 artificial intelligence; fundus disease; ophthalmology of traditional Chinese medicine; diabetic retinopathy; retinopathy of prematurity

科學家艾伦·纽厄尔和奥利弗·塞弗里奇在1955年分别提出了下棋与计算机模式识别的研究，人工智能（artificial intelligence， AI）雏形首次出现[1]。AI发展迅速，截至2018年，中国AI市值规模已达238.2亿元。AI正在惠及人类生活的方方面面，也推动了医疗行业的迅速发展。2017年，中山大学眼科中心建立了白内障AI平台，公布了眼科AI诊断和治疗系统的消息，从此开启了中国眼科界AI的新纪元。在此基础上，眼底疾病的AI诊断和治疗也在悄然崛起。利用AI技术对部分眼底疾病进行早期筛查能够延缓病程进展，有效地降低致盲率，故本文以糖尿病视网膜病变（diabetic retinopathy， DR）、早产儿视网膜病变（retinopathy of prematurity， ROP）为例，简述AI时代下眼底病变筛查方式的进展。

1 眼底病变筛查与预测的重要性