近年来，眼科人工智能(artificial intelligence，AI)迅猛发展，眼底影像因易获取及其丰富的生物信息成为研究热点，眼底影像的AI分析在眼底影像分析中的应用不断深入、拓展。目前，关于糖尿病性视网膜病变(diabetic retinopathy，DR)、年龄相关性黄斑变性(age-related macular degeneration，AMD)、青光眼等常见眼底疾病的临床筛查、诊断和预测已有较多AI研究，相关成果已逐步应用于临床实践。除眼科疾病以外，探究眼底特征与全身各种疾病之间的关系并据此研发AI诊断系统已经成为当下的又一热门研究领域。AI应用于眼底影像分析将改善医疗资源紧缺、诊断效率低下的情况，为多种疾病的筛查和诊断开辟“新赛道”。未来眼底影像AI分析的研究应着眼于多种眼底疾病的智能性、全面性诊断，对复杂性疾病进行综合性的辅助诊断；注重整合标准化、高质量的数据资源，提高算法性能、设计贴合临床的研究方案。
In recent years, artificial intelligence (AI) in ophthalmology has developed rapidly. Fundus image has become a research hotspot due to its easy access and rich biological information. The application of AI analysis in fundus image is under continuous development and exploration. At present, there have been many AI studies on clinical screening, diagnosis and prediction of common fundus diseases such as diabetic retinopathy (DR), age-related macular degeneration (AMD), and glaucoma, and related achievements have been gradually applied in clinical practice. In addition to ophthalmic diseases, exploring the relationship between fundus features and various diseases and developing AI diagnostic systems based on this has become another popular research field. The application of AI in fundus image analysis will improve the shortage of medical resources and low diagnostic efficiency, and open up a “new track” for screening and diagnosis of various diseases. In the future, research on AI analysis of fundus image should focus on the intelligent and comprehensive diagnosis of multiple fundus diseases, and comprehensive auxiliary diagnosis of complex diseases, and lays emphasis on the integration of standardized and high-quality data resources, improve algorithm performance, and design clinically appropriate research program.
传统的眼底手术要求眼科医生具备精细的操作技术，但即便拥有再精湛的操作技术，眼底手术还是存在很大的风险性。因此，为了减少手术风险，提高手术质量，对传统眼底手术进行改进是十分必要的。近年来，在我国对于人工智能产业的大力支持之下，应用于各类行业的机器人随之诞生。机器人辅助系统(robot auxiliary system，RAS)在医学领域，特别是眼科学中应用广泛。对近几年RAS应用于眼底手术的案例进行整理总结，并将RAS参与的眼底手术以及传统的眼底手术进行对比，可以发现RAS在眼底手术中的应用可以显著提高手术效率，并降低手术风险。未来RAS的发展趋势可能着重聚焦于与深度学习算法的紧密结合。通过算法对手术中的视野图像进行预测、优化，从而让高精度的眼底手术更加高效、安全。
Traditional fundus surgery requires ophthalmologists to be equipped with sophisticated operating techniques, but even with the most sophisticated operating techniques, fundus surgery still has great risks. Therefore, in order to reduce the risk of surgery and improve the quality of surgery, it is very necessary to improve the traditional fundus surgery. In recent years, with China’s strong support for the artificial intelligence industry, robots used in various industries have been born. Robot auxiliary system (RAS) is widely used in the medical field, especially in ophthalmology. By summarizing the cases of fundus surgery with RAS in recent years and comparing the fundus surgery involving RAS with traditional fundus surgery, it can be found that the application of RAS in fundus surgery can significantly improve the efficiency of surgery and reduce the risk of surgery. The future development trend of RAS may focus on the close integration with deep learning algorithms, which can predict and optimize the field of view images during surgery so that high-precision fundus surgery can be more efficient and safer.
人工智能是对人类智能的模拟和拓展。基于深度学习的人工智能可以很好地利用图像的内在特征，如轮廓、框架等，来分析图像。研究人员通常利用图像来诊断眼底病，因此将人工智能应用于眼底检查是有意义的。在眼科领域，人工智能通过分析光学相干断层扫描图像、眼底照片和超宽视野图像，已经在检测多种眼底疾病上取得了类似医生的性能。它也已经被广泛应用于疾病进展预测。然而，人工智能在眼科的应用也存在一些潜在的挑战，黑盒问题是其中之一。研究人员致力于开发更多的可解释的深度学习系统，并确认其临床可行性。人工智能在最流行的眼底病中的最新应用、可能遇到的挑战以及未来的道路将一一阐述。
Artificial intelligence (AI) is about simulating and expanding human intelligence. AI based on deep learning (DL) can analyze images well by using their inherent features, such as outlines, frames and so on. As researchers generally diagnoses ocular fundus diseases by images, it makes sense to apply AI to fundus examination. In ophthalmology, AI has achieved doctor-like performance in detecting multiple ocular fundus diseases through optical coherence tomography (OCT) images, fundus photographs, and ultra-wide-field (UWF) images. It has also been widely used in disease progression prediction. Nonetheless, there are also some potential challenges with AI application in ophthalmology, one of which is the black-box problem. Researchers are devoted to developing more interpretable deep learning systems (DLS) and confirming their clinical feasibility. This review describes a summary of the state-of-the-art AI application in the most popular ocular fundus diseases, potential challenges and the path forward.
目的：依托最新的第5代移动通信技术(5th generation wireless systems，5G)，构建基于眼底图片的5G医疗眼科远程诊断平台，促进医疗资源上下贯通，提升基层服务能力及医疗服务体系整体效能。方法：基于5G时代医院的信息化发展战略，在海南省卫生健康委员会的资助与指导下，中山大学中山眼科中心海南眼科医院与中国联通通信集团海南有限公司等进行跨行业、多学科的技术力量研究开发，构建5G条件下的平台建设模块和技术路线，确定远程眼科诊断流程，并在海南省内多地区应用。结果：远程诊断平台运行良好。2020年12月至2021年11月，本研究共在海南省17个地区的186个卫生院中开展，共收集1561例患者眼底病图片数据，筛查阳性例数为185例，检出眼底病总阳性率为11.9%。其中有42例需要转诊治疗，转诊率为23%；143例不需要转诊治疗，非转诊率为77%。在1561例眼底图像中，采集异常的眼底图像有490例。排除490例异常眼底图像后，辅助诊断系统与人工诊断结果有1 002张眼底图像诊断相同，69张眼底图像诊断不同，其辅助诊断系统准确率为93.3%。结论：5G移动通信与远程医学影像结合，运用互联网科技催生新型医疗生产力，提高卫生经济的质量和效率，是医疗领域探索5G应用场景的一项应用典范。
Objective: Relying on the latest 5th generation wireless systems (5G), a remote primary ophthalmology care diagnosis platform based on fundus images was constructed in order to promote the connectivity of medical resources and improve the primary health service capabilities and the overall effectiveness of the medical service system. Methods: Based on the 5G informatization development strategy of hospitals, and under the funding and guidance of the Hainan Provincial Health Commission, the Hainan Eye Hospital of Zhongshan Ophthalmic Center and China Unicom Communications Group Hainan Co., Ltd. conducted a cross-industry, multi-disciplinary technical research. To build platform construction modules and technical routes under 5G networks, present the remote ophthalmological diagnosis process, and apply it in many regions in Hainan Province. Results: The performance of the remote diagnosis platform is well. From December 2020 to November 2021, this study was carried out in 186 health centers in 17 regions of Hainan Province. A total of 1 561 patients with fundus disease image data were collected. The number of positive screening fundus disease cases was 185. The total positive rate was 11.9%. Among them, 42 cases required referral for treatment, with a referral rate of 23%, and 143 cases did not require referral for treatment, with a non-referral rate of 77%. Among 1 561 cases of fundus images, 490 fundus images were excluded due to abnormal quality. Compared the results of the diagnosis platform system with manual diagnosis, 1 002 fundus images were identical, and 69 fundus images were different in diagnosis. The accuracy of the auxiliary diagnosis system was 93.3%. Conclusions: The collaboration of 5G mobile communication and telemedicine imaging, combined with internet technology to promote new medical productivity, improve quality and efficiency of the health economy. This study is an application model for exploring 5G application scenarios in the medical field.
剥脱综合征(exfoliation syndrome,XFS)以眼内异常纤维样物质沉积为特征，临床典型表现为裂隙灯下瞳孔缘和(或)晶状体前囊膜存在灰白色粉末状的剥脱物(exfoliation material,XFM)。XFM可阻塞小梁网引起剥脱性青光眼(exfoliaiton glaucoma,XFG)，并可通过房水循环进入血液，引起血管性损害。眼底病变视力损伤通常不可逆，XFM可进入眼底微血管及毛细血管，引起眼底结构和血管异常。基于光学相干断层成像技术的光学相干断层扫描(optical coherence tomography,OCT)及光学相干断层扫描血管成像(optical coherence tomography angiography,OCTA)以实时、非侵入性、高分辨率等优势，已广泛应用于眼底组织结构及血管病变检查。文章对XFS眼底病变在OCT和OCTA上的表现进行综述。
Exfoliation syndrome (XFS) was characterized by the abnormal deposition of the fber-like material intraocularly, and manifested as white or gray, powdery exfoliation material (XFM) on the pupillary border and (or) anterior lens capsule under slit lamp microscopy. XFM could obstruct the trabecular meshwork and cause exfoliation glaucoma (XFG). In addition, XFM that entered aqueous humor circulation could enter bloodstream and result in vascular damage. XFM could enter ocular fundus microvascular and capillary vessels, causing abnormalities of fundus structures and vessels. Optical coherence tomography (OCT) and optical coherence tomography angiography (OCTA), which were based on optical coherence tomography technology, had the advantages of real-time, non-intrusive and high resolution, et al. OCT and OCTA were widely used in detection of fundus structural and vascular abnormalities. Tis study was to review the fundus lesion of XFS on OCT and OCTA.
目的：探讨眼底外科医护一体组团合作模式在管理眼底病日间手术患者实践效果。方法：选取2022年1—6月进行日间手术的582例眼底病患者为对照组，2023年1—6月进行日间手术的633例眼底病患者为研究组，对照组实施责任制整体护理，研究组采取实施医护一体组团合作管理模式进行全流程患者管理。使用χ2检验和t检验比较两组患者围术期护理知识健康教育知晓度、满意度、出院24 h内眼科急症就诊率、出院24 h内随访率、护士职业获益感的差异。结果：研究组患者健康教育知晓度高于对照组，组间比较差异具有统计学意义(t=–18.47，P＜0.05)；研究组患者满意度高于对照组，组间比较差异具有统计学意义(t=–4.005，P＜0.05)；研究组患者出院24 h内随访率为100%，对照组为98.1%，两组比较差异有统计学意义(χ2=12.073，P＜0.05)；研究组患者出院24 h内眼科急症就诊率0.94%，对照组为1.89%，组间比较差异无统计学意义(χ2=1.951，P=0.222)；实施后护士职业获益感分值高于实施前，组间比较差异具有统计学意义(t=–6.637，P＜0.001)。结论：医护一体组团合作管理模式进行眼底外科日间手术患者围术期的全流程管理，改善患者就医感受，提升患者就医体验，提高眼底外专科日间手术患者的依从性，保障患者的安全，提高护士职业获益感。
Objective: To investigate the practical effect of the integrated group cooperation model in managing patients with fundus diseases in day surgery. Methods: 582 patients with fundus disease who underwent day surgery from January to June in 2022 were included as the control group, and 633 patients with fundus disease who underwent day surgery from January to June in 2023 were selected as the study group. The control group implemented the overall responsibility nursing system, while the study group carried out the collaborative management model, integrating medical and nursing for the entire process of patient management. Chi-square test and T-test were used to compare the differences of perioperative nursing knowledge, health education awareness, satisfaction, emergency ophthalmological consultation rate within 24 hours of discharge, follow-up rate within 24 hours of discharge, and nurses' sense of professional benefit between the two groups. Results: The awareness of health education in the study group was higher than that in the control group, and there was significant statistically difference between two groups (t=–18.47, P<0.05). The satisfaction of patients in the study group was higher than that in the control group, and there was significant statistically difference between two groups (t=–4.005, P ＜ 0.05). The follow-up rate within 24 hours after discharge was 100% in the study group and 98.1% in the control group, and the difference was statistically significant (χ2=12.073, P<0.05). The incidence of ophthalmic emergencies within 24 hours of discharge in the study group was 0.94%, while in the control group it was 1.89%. There was no statistically significant difference between the two groups (χ2= 1.951, P=0.222). The perceived benefit score of nurses after implementation was higher than that before implementation, and there was statistically significant between two groups (t=–6.637, P<0.05). Conclusions: The medical and nursing integrated group cooperation management model is used to manage the entire perioperative process of patients undergoing day surgery in fundus surgery. This model can improve patients' medical experience, enhance their compliance with ophthalmic surgery, ensure their safety. At the same time, it can enhance the senses of professional benefits for nurses.
目的：评价欧堡Daytona 200度超广角激光扫描检眼镜检查近视患者眼底周边部视网膜病变的应用价值。方法：本研究为前瞻性病例研究，收集爱尔眼科医院要求行屈光手术的近视患者1 000例(2 000只眼)，分别进行小瞳下欧堡Daytona 200度超广角激光扫描检眼镜眼底检查和散瞳后三面镜检查，记录检查结果并进行比较分析。结果：通过欧堡Daytona 200度超广角激光扫描检眼镜检查发现有周边视网膜病变共230例(310只眼)，检出阳性率为15.50%；三面镜检查发现周边部视网膜病变共242例(322只眼)，检出阳性率为16.10%。两种检查方法对近视患者周边部视网膜病变检出阳性率具有很好的一致性(Kappa值0.8~1.0)。结论：欧堡Daytona 200度超广角成像系统为检查周边部视网膜病变提供了更省时高效的方法，在屈光手术前筛查视网膜周边部病变，具有广阔的临床应用前景。
Objectives: To evaluate the clinical value of peripheral retinal diseases in myopic patients examined by 200-degree ultra-wide field laser ophthalmoscope (Daytona). Methods: This was a prospective case-control study. We collected 1 000 myopic patients (2 000 eyes) who were scheduled to undergo refractive surgery in Aier Eye Hospital. They were examined by 200-degree ultra-wide field laser ophthalmoscope (Daytona) with non-mydriasis and three-mirror contact lens with mydriasis. The examination results were recorded and statistically analyzed. Results: A total of 230 cases (310 eyes) with peripheral retinopathy were found by 200-degree ultra-wide field laser ophthalmoscope (Daytona). The positive rate was 15.50%; 242 cases (322 eyes) with peripheral retinopathy were found by three- mirror contact lens, and the positive rate was 16.10%. The two methods were consistent in the detection of peripheral Retinopathy in myopic patients (the Kappa value is between 0.8 and 1.0). Conclusion: 200-degree ultra-wide field laser ophthalmoscope (Daytona) is an effective and rapid method for detecting peripheral retinopathy. It provides a broad clinical application prospects for peripheral retinopathy screening before refractive surgery.
目的：探讨应用数字化广域成像系统RetCam3行婴幼儿口服荧光素钠眼底血管荧光造影(fluorescein fundus angiography，FFA)的护理。方法：选择2018年8月至2019年12月在广州中山大学中山眼科中心小儿眼病综合科就诊的眼底疾病婴幼儿78例，应用RetCam3进行口服法FFA检查及护理，将护理要点进行总结。结果：所有患儿安全、顺利完成检查，检查过程中均未发生异常病情变化或与检查、药物相关的并发症。经FFA确诊家族性渗出性玻璃体视网膜病变(familial exudative vitreoretinopathy ，FEVR)26例，早产儿视网膜病变(retinopathy of prematurity ，ROP)23例，色素失禁症患者6例；玻璃体积血患者3例；视网膜母细胞瘤患者3例；牵牛花综合征患者1例；视网膜色素变性患者3例；弓蛔虫眼病患者1例；原始永存玻璃体患者2例；不明原因眼底病变患者5例，单眼视网膜皱襞患者1例，先天性小眼球患者1例，巨细胞病毒感染患者1例，先天性黄斑发育不良患者1例；Coats病患者1例。结论：应用RetCam3行婴幼儿口服法FFA是一种安全、有效的检查方法。规范、恰当的护理配合能够保证检查准确、顺利地完成。
Objective: To share the nursing experience of RetCam3 ultra-widefield oral fluorescein fundus angiography (FFA) in infants with fundus diseases. Methods: Seventy-eight infants with fundus diseases admitted to General Department of Pediatric Ophthalmology in Zhongshan Ophthalmic Center, Sun Yat-sen University from August 2018 to December 2019 were recruited. Oral FFA was carried out using the 130-degree lens of RetCam3, and the key points of nursing were summarized. Results: No complications related to the examination and drugs occurred after oral FFA with an appropriate nursing manner. FFA confirmed 26 cases of familial exudative vitreoretinopathy,23 cases of retinopathy of prematurity and 6 cases of pigment incontinence. Vitreous hematoma was observed in 3 patients, retinoblastoma in 3 patients, Morning Glory syndrome in 1 patient, retinitis pigmentosa in 3 patients,Ascaris lumbricoides eye disease in 1 case, original permanent vitreous body in 2 patients, unexplained fundus lesions in 5 patients, monocular retinal fold in 1 patient, congenital micro-eyeball in 1 patient, cytomegalovirus infection in 1 patient, congenital macular dysplasia 1 patient and Coats disease in 1 patient. Conclusion: Oral FFA with RetCam3 is an effective and safe detection method for infants. Standard and proper nursing can ensure the examination can be performed accurately and smoothly.
目的：借助于人工智能(artificial intelligence,AI)眼底筛查远程接转诊系统，探索“患者-社区-医院”远程筛查模式，推进眼科分级诊疗和双向转诊实施，为地市级医疗机构开展眼底疾病人工智能筛查工作提供一定的经验借鉴。方法：通过AI辅助远程筛查基层医疗机构的4886例患者，完成眼科检查并经AI初判、人工复核形成眼底诊断结论。通过医联体和专科联盟模式，对基层医疗机构的4886例患者的AI诊断系统结果和上级医师审核结果进行对照分析，分析AI诊断系统在眼科常见病种筛查中的推广应用的可信度和可行性。结果：AI检出DR的灵敏度为94.70%，特异度96.06%；DME的灵敏度96.43%，特异度96.55%；AMD的灵敏度77.55%，特异度95.74%；同时，其在病理性近视、白内障、青光眼等常见病种眼底筛查中也有一定作用。结论：AI辅助远程筛查系统对于绝大多数眼底疾病有较高的敏感性和特异性，适用于眼底疾病的筛查工作，利于基层医院或社区医院对于眼底疾病的初步诊断，落实眼科分级诊疗，有借鉴推广意义。
Objective: With the help of artificial intelligence (AI) based fundus screening remote referral telemedicine system,it enables us to explore the remote screening mode of patient-community-hospital, and promote the two-way referral and ophthalmic graded diagnosis. This investigation provides certain practice experiences for prefecture-level medical institutions to carry out AI screening for fundus diseases. Methods: Ophthalmologic examination was performed on 4,886 patients in primary medical institutions through AI-aided remote screening, and the final fundus diagnosis conclusion was formed after AI preliminary judgment and manual review. Through the Medical Consortium and specialty alliance model, the results of the AI diagnosis system and the audit results of superior physicians for 4 886 patients in primary care institutions were compared and analyzed, and the credibility and feasibility of the AI diagnosis system application in the screening of common ophthalmic diseases were discussed. Results: The sensitivity and specificity of AI detection of diabetic retinopathy were 94.70% and 96.06%, respectively. In the diabetic macular edema classification, the sensitivity and specificity were 96.43% and 96.55%, respectively. In the age-related macular degeneration classification, the sensitivity and specificity were 77.55% and 95.74%, respectively. Meanwhile, it also plays a role in screening common fundus diseases such as pathological myopia, cataract and glaucoma. Conclusion: The AI-aided remote screening system has high sensitivity and specificity for most of fundus diseases, indicating it is promising for fundus diseases screening in primary medical institutions. It is conducive for primary hospitals or community hospitals to carry out the initial diagnosis of fundus diseases, as well as the implementation of graded diagnosis and treatment of ophthalmology, which has reference and promotion significance.
目的：探索基于眼底彩照和人工智能构建冠心病智能诊断系统的可行性。方法：于2013—2014年收集广东省人民医院530例患者共2117张眼底彩照，其中冠心病217例共909张眼底彩照。根据患者有无冠心病的情况进行标记，使用Inception-V3深度卷积神经网络训练人工智能模型，随后使用验证数据判断模型的准确率。计算深度卷积网络模型的准确性、一致率、敏感性、特异性和受试者工作特性曲线下面积(area under the curve，AUC)。结果：在2117张眼底彩照中，1903张用于模型训练，214张用于模型的性能评估。在测试集中，该算法的准确性为98.1%，一致率为98.6%，敏感性为100.0%，特异性为96.7%，AUC为0.988(95%CI：0.974~1.000)。结论：眼底彩照联合人工智能技术可精准判定冠心病，该模型具备较高的敏感性和特异性，但须进一步增加样本量，使用大样本量数据验证该模型，排除过拟合的可能性。
Objective: To explore the feasibility of developing a deep learning algorithm for detecting coronary heart diseases based on fundus color photography and artificial intelligence (AI). Methods: A total of 2 117 fundus color photographs were taken from 530 patients in Guangdong Provincial People’s Hospital from 2013 to 2014,including 909 fundus color photographs from 217 patients with coronary heart disease (CHD). According to whether the patient had coronary heart disease or not, the Inception-V3 depth convolution neural network was used to train the deep learning model, and then the validation data were used to judge the accuracy of the model. The accuracy, consistency rate, sensitivity and specificity of the deep convolution network model and the area under the working characteristic curve (AUC) were calculated. Results: Among the 2 117 fundus color photographs, 1 903 were used for model training, and 214 were used to test the accuracy of the model. In the test dataset, the accuracy of the algorithm was 98.1%, the consistency rate was 98.6%, the sensitivity was 100.0%, and the specificity was 96.7%. The AUC was 0.988 (95% CI, 0.974–1.000). Conclusion: The combination of fundus color photography and artificial intelligence can achieve the accurate diagnosis of the coronary heart disease, and the model has high sensitivity and specificity. However, future studies are warranted to validate our model and exclude the possibility of over-fitting.
