您的位置: 首页 > 2021年12月 第36卷 第12期 > 文字全文
2023年7月 第38卷 第7期11
目录

人工智能在近视防控与治疗中的应用进展

Advances in the application of artificial intelligence in the prevention, control and treatment of myopia

来源期刊: 眼科学报 | 2021年12月 第36卷 第12期 965-971 发布时间:2021-12 收稿时间:2023/5/9 8:47:35 阅读量:3946
作者:
关键词:
人工智能近视机器学习防控治疗
artificial intelligence myopia machine learning prevention and control treatment
DOI:
10.3978/j.issn.1000-4432.2021.09.03
近视是危害儿童青少年视力最常见的眼部疾病,高度近视对视功能造成极大的威胁。近年来,我国近视发病率逐年升高,对近视筛查与防控的需求也不断增加,随着人工智能理论与技术的不断发展与成熟,可以辅助眼科医生进行近视筛查、诊断与治疗。本文将简要介绍人工智能在近视的筛查、预测、检测、病理性近视以及角膜屈光手术中的应用,浅谈了目前人工智能在研究中存在的可比度较低、影像要求较高、可解释性较低及隐私保护等问题,并展望人工智能在近视相关领域的应用前景。
Myopia is the most common ocular disease that harms the vision of children and adolescents. High myopia poses a great threat to visual function. The incidence of myopia in China has been increasing in recent years, and the demand for myopia screening, prevention and control has also expanded. With the continuous development of artificial intelligence theory and technology, Artificial intelligence can assist ophthalmologists in myopia screening, diagnosis and treatment. This review will briefly introduce artificial intelligence in the screening, prediction, and detection of myopia; also, the application in pathological myopia and corneal refractive surgery. This review will discuss some problems of current artificial intelligence research, such as low comparability, high image requirements, low interpretability, privacy protection, and the application prospects of artificial intelligence in myopia.
近视是由于角膜曲率过高、眼轴过长等,导致远处物体在视网膜前聚焦,使远距离视力模糊[1],是影响人类视功能主要的原因[2]。我国学龄儿童近视患病率逐年增长,影响青少年学习生活、生长发育和身心健康。近视的快速增长已成为全球公共卫生领域重点关注的问题[3]。目前全球约有超过14亿人患有近视,占总人口的22.9%,1.6亿人患有高度近视,占总人口的2.7%;到2050年,全世界近视人口总数预计会超过47亿[4]。根据国家卫健委发布的最新数据[5],我国儿童青少年总体近视率为52.7%,其中6岁儿童为14.3%,小学生为35.6%,初中生为71.1%,高中生为80.5%,大学生总体发生率为90%,高于世界平均水平。
随着我国近视发病率的不断攀升,近视相关的早期筛查、诊断、视力矫正的需求也逐年升高,传统医疗手段已难以满足现实近视防控的需要,人工智能技术为解决这一困境提供了新的思路。近年来,人工智能算法和核心技术快速发展,在医疗卫生领域中的应用越来越广泛[6],可以运用大数据技术对临床影像资料和相关数据进行汇集,利用计算机强大的算法分析、挖掘数据,辅助医生或直接参与临床决策[7]
目前,人工智能在眼科领域的运用十分广泛,主要集中于眼科患者资料的分析和处理,对于白内障、青光眼、糖尿病视网膜病变、年龄相关性黄斑变性、早产儿视网膜病变等疾病的筛查与诊断具有重要意义[8-12]。在近视防控领域,人工智能的应用仍处于探索阶段,依托大数据、人工智能和精准医疗,进行近视防控的六维度量化监测评估的新模式,即通过遗传度、屈光度、用眼度、裂变度、病理度、干预度等6个维度进行近视防控,是未来近视精准化、个性化防控的新方向[13]

1 人工智能在近视筛查、预测与监测中的应用

1.1 近视筛查

目前我国近视筛查的最主要手段为在每年中小学的体检中,由受过专业训练的护士或技术人员进入学校进行裸眼视力检查,再将视力异常的儿童转诊至医院进一步验光以明确诊断。最新研究[14]显示:裸眼视力筛查的平均敏感性为59.71%,平均特异性为89.74%。校园筛查存在漏诊率高、检查结果反馈不及时、人力成本高、偏远地区覆盖率较低等问题[15],使许多近视患儿的病情延误,屈光不正没有得到及时的诊断和治疗。Yang等[16]用2 350幅用通常照相机拍摄的6~18岁患儿的眼球外观图像,训练深度学习系统通过眼球外观识别异常屈光状态,结果显示敏感性为81.3%,特异性为86.42%,AUC为0.927,敏感性远高于传统学校传统裸眼视力检测手段。此外,新的近视筛查技术也在逐步诞生,为未来人工智能在相关领域的落地应用提供基础,如Jaeb Visual Acuity Screener[17]——一种公开免费的近视筛查软件,家长可在家中用家用电脑对儿童进行近视筛查;SVone[18]——一种可与市面上通用的智能手机连接的外接设备,可随时进行屈光异常筛查,这些技术可以通过移动智能设备,远程实时监控青少年的屈光状态,有利于大规模推广与普及近视筛查,有效降低时间成本和人力成本,对近视的公共卫生防控具有重要意义。

1.2 近视预测

除对青少年近视进行早期筛查诊断外,基于现有资料,对未来学龄儿童近视发展变化进行预测,也是当下人工智能研究的热点。Lin等[19]对国内多中心来源的10年内超过68万份电子病历数据进行分析,训练人工智能算法,用以预测患者未来是否会进展为高度近视,结果显示:3年内预测准确性超过了90%,8年内预测准确性超过了80%。该研究首次运用大数据和人工智能提供了高度近视高准确性的临床预测模型,将高度近视的早期识别诊断提前了近8年,对提前做好高度近视的预防与筛查,减少高度近视远期并发症以及个体近视精准防控具有重要意义。Yang等[20]基于机器学习理论,对影响近视形成的因素如是否佩戴框架眼镜、室内活动时间、户外活动时间、眼轴角膜曲率、饮食、行为习惯等进行分析研究,以SVM模型为基础,提供了一种新型的青少年近视预测模型,为青少年近视预测提供了一种新思路。唐涛等[21]利用机器学习模型研究眼轴增加量与等效球镜度增加的对应关系,结果显示:眼轴增长1 mm所需要的时间跨度越大,对应近视增长度数越小,并给出了预测模型,方便眼科医师及视光学医师通过眼轴增加判断青少年近视变化情况。Varadarajan等[22]应用226 870幅眼底彩超训练“attention”模型,提取眼底图像特征,预测现有屈光度数,取得了较高的准确率。

1.3 近视监测

人工智能的进展促进了物联网时代智能穿戴设备的应用,新型智能穿戴设备可以对儿童青少年用眼姿势、习惯进行实时检测。其中代表性设备是中南大学爱尔眼科学院研发的“云夹”[23],其能够实时、全面记录近视工作距离、周边环境光照水平以及有无紫外线等因素,并将信息上传至云端由人工智能进行分析,对不良用眼习惯进行实时提醒和纠正。研究[24]表明:佩戴云夹可以有效防止不良姿势和近距离读写行为,并且在停止佩戴后依然可以维持一段时间,有效减缓近视的形成和进展。居玲等[25]为及时纠正儿童青少年不良用眼习惯,开发出“AI眼宝”APP,可以自动监测青少年读写姿势、光线明暗、用眼时间,利用人工智能进行语音提示、自动调节,可有效减缓近视进展。线下移动智能穿戴设备收集数据,与线上云计算近视防控大数据平台相结合,实时共享家庭、学校和医院的近视防控信息,有利于实时监测近视防控效果,提高儿童青少年近视早期干预成效[26]

2 人工智能在病理性近视中的应用

2.1 病理性近视的定义

病理性近视目前比较公认的定义为高度近视同时伴有巩膜、脉络膜、RPE病理性改变和视力损伤。2015年,Ohno-Matsui等[27]将病理性近视黄斑病变分为5级:无视网膜退行性病变为0级,豹纹状眼底为1级,弥漫性脉络膜视网膜萎缩为2级,斑块状脉络膜视网膜萎缩为3级,黄斑萎缩为4级,以及另外3个附加病变:漆裂纹、脉络膜新生血管和Fuchs斑。在此标准中,2级以上或具有至少1个附加病变即可诊断为病理性近视。
各种眼底病变如黄斑劈裂、视网膜脱离、脉络膜新生血管等是高度近视的常见并发症[28]。目前,病理性近视的标准分级、发病机制、预防策略和治疗方法等都还有待研究[29]。缺乏标准分级使得不同流行病学调查和临床实验的结果之间难以进行比较,没有统一的标准,也给基于大数据机器学习的人工智能对病理性近视眼底病变诊疗造成了困难。

2.2 人工智能对病理性近视的辅助诊断

病理性近视的诊断高度依赖于眼底病变的影像分析,近年来,人工智能在医学影像采集和识别领域取得了突破性的进展。Zapata团队[30]开发的Optretina远程医疗平台,眼底影像采集后,由眼科专科医生进行标记,作为训练集合,训练后,Optretina可对眼底影像进行分类并诊断黄斑病变,敏感性达到97.7%,特异性达到92.4%,准确性可达96%。Li等[31]收集了1 048位高度近视患者共计5 505幅眼底光学相干断层扫描(optical coherence tomography,OCT)图像,采用InceptionResnet V 2架构训练卷积神经网络(convolutional neural network,CNN)模型,用以识别4种在高度近视中威胁视力的眼底并发症,即视网膜劈裂、黄斑裂孔、视网膜脱离和脉络膜新生血管形成。训练后的系统表现良好,敏感性略胜于视网膜病专科医师,特异性超过90%。Hemelings等[32]同样基于CNN模型,训练人工智能对视网膜影像进行自动诊断和分类,同时将病变部位(视盘、视网膜萎缩、视网膜脱离)在影像上进行标注。
人工智能还可对眼底影像中的生理病理结构进行自动化分割和数据分析[33]。Fu等[34]搭建多标签深度学习网络和极性转换系统,对眼底影像中的视杯视盘进行了精准的自动化分割。Jiang等[35]基于区域的CNN研发了JointRCNN算法,联合特征提取模块、区域提议网络、视盘关注模块、功能丢失计算以及分割模块,完成对视盘周边病灶的自动化检测,测定病灶范围。Dodo等[36]利用模糊直方图增生法和图切算法,在OCT影像上跨过8个边界将视网膜分为7层并注释。Wu等[37]提出了基于深度学习算法NFN+算法,具有新型的级联设计和网络间连接,能够比现有模型更准确地在眼底图像上标注视网膜血管,辅助眼科医师进行诊断和治疗。Tan等[38]基于超过225 000幅视网膜图像训练深度学习算法,用以识别近视性黄斑变性及高度近视,进行了回顾性多国多队列的研究,其诊断AUC可达0.969,表现超过6名眼科专家。Sogawa等[39]运用SS-OCT影像训练CNN,用以识别无黄斑病变的正常眼底与有近视性黄斑病变的眼底图像,AUC可达0.970,有较高的准确性。
在人工智能的辅助下,眼科医师可以更快更准确地完成对眼底病变的识别与分析,有效减少读片时间,极大提高眼底影像分析效率。利用人工智能进行病理性近视的初步诊断与筛查,也可以减轻眼科医师的工作量、减少病理性近视筛查的人力物力成本。

3 人工智能在近视矫正与治疗中的应用

目前,治疗近视的主要手段包括佩戴框架眼镜、角膜塑形镜、角膜屈光手术等[40-42]。人工智能在相关领域的研究也有许多进展,如王凯团队[43]依据我国青少年既往的角膜地形图和屈光数据,应用人工智能研发出无接触、个性化角膜塑形镜免试戴配镜法,准确率可达90%以上。
近年来,随着我国近视发病率不断增加,对近视屈光手术的需求也不断增加,新型屈光手术技术不断应运而生[44],对屈光手术近视术前筛查以及术后随访的工作量急剧增加。人工智能对影像资料强大的识别与解析能力,可以对角膜影像进行自动化、客观、有效的分析,确定适合手术的患者,在屈光手术术前筛查及术后并发症监测领域具有广阔的应用前景,但目前的研究仍然在小样本范围内进行,未来的实际运用需要进一步探索[45]
圆锥角膜是引起高度近视的常见病因之一,也是角膜屈光手术的禁忌证之一,其诊断高度依赖影像资料[46]。Lavric团队[47]基于CNN开发了KeratoDetect算法,经过训练,可以从角膜断层显像中高精度地检测圆锥角膜,准确率可达99.33%,并能够监测疾病的进展。Xie等[48]基于深度学习架构研发的PIRSS系统,对准备接受角膜屈光手术的患者进行筛查,排除存在术后继发角膜扩张风险以及有圆锥角膜的患者,该团队在两年半的时间里,搜集了1 385个准备接受角膜屈光手术的患者的角膜层析影像数据,PIRSS系统用6 465个角膜影像进行训练,对怀疑术后继发角膜扩张的诊断敏感性可达80%,对诊断早期圆锥角膜的敏感性可达95%,诊断总体准确率达到95%,AUC为0.99,通过人工智能进行术前筛查,能够有效减少角膜屈光手术的术后并发症,降低手术风险。
Cui等[49]分析了机器学习技术用于预测全飞秒激光近视手术列线图的结果,研究表明:机器学习安全性预测与眼科医生相当,疗效预测优于眼科医生,但在高度近视与散光的预测上不如专业的眼科医生。Achiron等[50]应用统计分类算法基于17 592个患者的38个临床参数训练机器学习,可用于近视屈光手术前的风险评估,支持临床决策,帮助患者通过手术获益。未来人工智能在近视屈光手术术前筛查、术后预后预测及并发症监测等方面有着广泛的应用前景。

4 问题

4.1 可重复性和可比性较低

目前几乎所有人工智能和机器学习的研究,都是基于不同数据集的分析和训练,不同数据集的影像资料分辨率、成色等都有一定的差异,很多团队的数据库和源代码都没有公开,使得实验可重复性和不同算法之间的对比研究变得十分困难。也正因如此,人工智能算法和系统的研究与实际临床应用落地之间还存在着较大的距离。为解决这一问题,在严格遵守患者隐私保护的前提下,可鼓励研究团队公开发表数据集,并由专业机构集中管理,将数据集储存于广泛应用的机器学习数据库中,供研究团队使用,以增加后续研究的可重复性和可比性[51]。同时也可以选择添加扩展程序,Wang等[52]为解决人工智能在分析不同机构、不同扫描程序提供的眼底影像存在障碍的问题,提出了一种新颖的基于补丁的输出空间对抗学习框架,使人工智能经过训练,对新数据集产生适应,逐步兼容。

4.2 影像资料质量要求高

人工智能的应用往往高度依赖眼部的影像资料,因而对影响资料的质量要求较高。对比度及像素较低的影像资料会使人工智能的判读能力降低,出现高特异性低敏感性的结果,典型的病例依然可以被正确判读,但会出现一定数量的假阳性结果。为解决这一问题,Zhang等[53]研发的超广角眼底筛查系统DeepUWF,创新地引入了6种眼底图像预处理技术:直方图均衡、自适应直方图均衡(adaptive histogram equalization,AHE)、强度缩放、伽马校正、S形调整和有限对比度AHE,有效提高了人工智能神经网络的学习能力,显著增加了实验结果的敏感性。

4.3 可解释性较低

尽管人工智能具有极其强大的分析、学习、预测能力,但目前大多数机器学习算法都对其诊断过程缺乏解释能力,其学习过程及决策过程的具体步骤是未知的(即“黑箱”)。人工智能的训练过程依赖于建立输入和输出结果间的联系。因此,部分算法并不依据影像资料中的病理信息进行诊断,而是综合了图片上其他混杂的特征,使其专业信服度降低。对“黑箱”进行拆解,决策步骤进一步细分,有助于提高机器学习算法的可解释性,也有助于临床医生通过人工智能,学习新的临床思维,进一步提高临床诊断能力。

4.4 数据监管与隐私保护尚不成熟

尽管人工智能在近视防控与治疗领域、乃至整个医疗领域都有着相当广阔的应用前景,人工智能方便、高效、快捷,成本效益比极佳,但这一切效率的前提来自于对众多患者真实临床数据的收集,由此不可避免地带来患者隐私泄露、数据滥用、决策公平等问题,目前我国在人工智能相关隐私问题的法律法规尚不够完善,不同机构的患者数据保密工作水平参差不齐,有关部门可加强相关领域的监管,研究机构应切实履行保护患者隐私的义务,以减少相关的伦理与法律问题,促进人工智能研究领域的发展。

4.5 研究范围较为局限,交叉研究较少

目前,人工智能在近视问题的研究多局限在对影像资料的处理,具体表现为:疾病的筛查与预测、病理性近视的诊断与分类、以及近视矫正与屈光手术治疗等领域,但人工智能联合基因组学、蛋白组学、环境科学以及人文社会科学等综合学科进行近视防控的研究还相对较少,一方面是人工智能技术近年来发展过于迅猛,交叉领域研究需要时间;另一方面,跨学科的综合研究需要有关部门牵头合作,整合资源。不可否认的是,要实现人工智能对青少年近视的精准防控,需要进一步拓展目前人工智能的研究领域,注重人工智能结合其他领域的跨学科探索。

5 结语

由于眼科学自身学科的特点,临床诊断依赖对影像资料的解析,人工智能强大的图像分析能力使其在眼科领域具有较大的应用前景。目前,有关人工智能在近视防控中的应用大多还处于试验阶段,随着临床研究逐步展开,未来人工智能在真实临床场景中诊治的准确性与稳定性将逐步提高。同时,伴随着大数据、5G技术与物联网技术的快速发展,更多的可穿戴智能设备和APP能实时可靠地监测儿童青少年的用眼习惯、发现屈光异常,监控近视进展,为近视的早期干预与预防提供条件。同时,智能设备所收集的海量数据,可在云端收集储存,建立数据库,为今后的人工智能训练提供新的素材;也可进行大数据分析,帮助研究人员进一步了解近视的流行病学及发病机制,为今后人工智能近视精准防控的大规模深度推广打下基础。同时,伴随着基因检测和各地电子病历系统的逐步普及,未来人工智能可逐步整合患者的综合信息,为近视的个体化防控提供可能。利用人工智能可大幅度降低近视筛查的时间成本及人力物力成本,可有效减缓高度近视、病理性近视进展,减少眼底病变的产生,降低近视致盲的发病率,减少由高度近视带来的公共卫生负担。

开放获取声明

本文适用于知识共享许可协议(Creative Commons),允许第三方用户按照署名(BY)-非商业性使用(NC)-禁止演绎(ND)(CC BY-NC-ND)的方式共享,即允许第三方对本刊发表的文章进行复制、发行、展览、表演、放映、广播或通过信息网络向公众传播,但在这些过程中必须保留作者署名、仅限于非商业性目的、不得进行演绎创作。详情请访问:https://creativecommons.org/licenses/by-nc-nd/4.0/。
1、Baird PN, Saw SM, Lanca C, et al. Myopia[ J]. Nat Rev Dis Primers, 2020, 6(1): 99.Baird PN, Saw SM, Lanca C, et al. Myopia[ J]. Nat Rev Dis Primers, 2020, 6(1): 99.
2、Modjtahedi BS, Ferris FL 3rd, Hunter DG, et al. Public health burden and potential interventions for myopia[ J]. Ophthalmology, 2018, 125(5): 628-630.Modjtahedi BS, Ferris FL 3rd, Hunter DG, et al. Public health burden and potential interventions for myopia[ J]. Ophthalmology, 2018, 125(5): 628-630.
3、Dolgin E. The myopia boom[ J]. Nature, 2015, 519(7543): 276-278.Dolgin E. The myopia boom[ J]. Nature, 2015, 519(7543): 276-278.
4、Holden BA, Fricke TR, Wilson DA, et al. Global prevalence of myopia and high myopia and temporal trends from 2000 through 2050[ J]. Ophthalmology, 2016, 123(5): 1036-1042.Holden BA, Fricke TR, Wilson DA, et al. Global prevalence of myopia and high myopia and temporal trends from 2000 through 2050[ J]. Ophthalmology, 2016, 123(5): 1036-1042.
5、国家卫生健康委宣传司. 国家卫生健康委员会2021年7月13 日新闻发布会文字实录[EB/OL]. [2021-07-13]. [2021-07-29]. http://www.nhc.gov.cn/xcs/s3574/202107/2fef24a3b77246fc9fb36d c8943af700.shtml
Department of Publicity, National Health Commission. Transcript of press Conference held by National Health Commission on July 13, 2021[EB/OL]. [2021-07-13]. [2021-07-29]. http://www.nhc.gov.cn/ xcs/s3574/202107/2fef24a3b77246fc9fb36dc8943af700.shtml
国家卫生健康委宣传司. 国家卫生健康委员会2021年7月13 日新闻发布会文字实录[EB/OL]. [2021-07-13]. [2021-07-29]. http://www.nhc.gov.cn/xcs/s3574/202107/2fef24a3b77246fc9fb36d c8943af700.shtml
Department of Publicity, National Health Commission. Transcript of press Conference held by National Health Commission on July 13, 2021[EB/OL]. [2021-07-13]. [2021-07-29]. http://www.nhc.gov.cn/ xcs/s3574/202107/2fef24a3b77246fc9fb36dc8943af700.shtml
6、Yu KH, Beam AL, Kohane IS, et al. Artificial intelligence in healthcare[ J]. Nat Biomed Eng, 2018, 2(10): 719-731.Yu KH, Beam AL, Kohane IS, et al. Artificial intelligence in healthcare[ J]. Nat Biomed Eng, 2018, 2(10): 719-731.
7、王晓刚, 王晓亮, 董静. 眼科人工智能应用研究进展[ J]. 中华眼 外伤职业眼病杂志, 2020, 42(11): 876-880.
WANG Xiaogang, WANG Xiaoliang, DONG Jing. Advances in the application of artificial intelligence in ophthalmology[ J]. Chinese Journal of Ocular Trauma and Occupational Eye Disease, 2020, 42(11): 876-880.
王晓刚, 王晓亮, 董静. 眼科人工智能应用研究进展[ J]. 中华眼 外伤职业眼病杂志, 2020, 42(11): 876-880.
WANG Xiaogang, WANG Xiaoliang, DONG Jing. Advances in the application of artificial intelligence in ophthalmology[ J]. Chinese Journal of Ocular Trauma and Occupational Eye Disease, 2020, 42(11): 876-880.
8、Xu X, Zhang L, Li J, et al. A hybrid global-local representation CNN model for automatic cataract grading[ J]. IEEE J Biomed Health Inform, 2020, 24(2): 556-567.Xu X, Zhang L, Li J, et al. A hybrid global-local representation CNN model for automatic cataract grading[ J]. IEEE J Biomed Health Inform, 2020, 24(2): 556-567.
9、Kim KE, Kim JM, Song JE, et al. Development and validation of a deep learning system for diagnosing glaucoma using optical coherence tomography[ J]. J Clin Med, 2020, 9(7): 2167.Kim KE, Kim JM, Song JE, et al. Development and validation of a deep learning system for diagnosing glaucoma using optical coherence tomography[ J]. J Clin Med, 2020, 9(7): 2167.
10、Gargeya R, Leng T. Automated identification of diabetic retinopathy using deep learning[ J]. Ophthalmology, 2017, 124(7): 962-969.Gargeya R, Leng T. Automated identification of diabetic retinopathy using deep learning[ J]. Ophthalmology, 2017, 124(7): 962-969.
11、Hwang DK, Hsu CC, Chang KJ, et al. Artificial intelligence-based decision-making for age-related macular degeneration[ J]. Theranostics, 2019, 9(1): 232-245.Hwang DK, Hsu CC, Chang KJ, et al. Artificial intelligence-based decision-making for age-related macular degeneration[ J]. Theranostics, 2019, 9(1): 232-245.
12、Campbell JP, Kim SJ, Brown JM, et al. Evaluation of a deep learningderived quantitative retinopathy of prematurity severity scale[ J]. Ophthalmology, 2021, 128(7): 1070-1076.Campbell JP, Kim SJ, Brown JM, et al. Evaluation of a deep learningderived quantitative retinopathy of prematurity severity scale[ J]. Ophthalmology, 2021, 128(7): 1070-1076.
13、徐捷, 徐亮. 近视防控的六维度评估及防控模式[ J]. 中华眼视光 学与视觉科学杂志, 2018, 20(3): 129-132.
XU Jie, XU Liang. Six-dimensional evaluation for myopia prevention and control[ J]. Chinese Journal of Optometry Ophthalmology and Visual Science, 2018, 20(3): 129-132.
徐捷, 徐亮. 近视防控的六维度评估及防控模式[ J]. 中华眼视光 学与视觉科学杂志, 2018, 20(3): 129-132.
XU Jie, XU Liang. Six-dimensional evaluation for myopia prevention and control[ J]. Chinese Journal of Optometry Ophthalmology and Visual Science, 2018, 20(3): 129-132.
14、Lin S, Ma Y, He X, et al. Using decision curve analysis to evaluate common strategies for myopia screening in school-aged children[ J]. Ophthalmic Epidemiol, 2019, 26(4): 286-294.Lin S, Ma Y, He X, et al. Using decision curve analysis to evaluate common strategies for myopia screening in school-aged children[ J]. Ophthalmic Epidemiol, 2019, 26(4): 286-294.
15、林浩添, 林铎儒. 我国未成年人视觉损害人工智能诊疗研究的 现状和建议[ J]. 中国临床新医学, 2020, 13(2): 111-114.
LIN Haotian, LIN Duoru. Current situation and suggestion of the research on artificial intelligence diagnosis and treatment of visual impairment of minors in China[ J]. Chinese Journal of New Clinical Medicine, 2020, 13(2): 111-114.
林浩添, 林铎儒. 我国未成年人视觉损害人工智能诊疗研究的 现状和建议[ J]. 中国临床新医学, 2020, 13(2): 111-114.
LIN Haotian, LIN Duoru. Current situation and suggestion of the research on artificial intelligence diagnosis and treatment of visual impairment of minors in China[ J]. Chinese Journal of New Clinical Medicine, 2020, 13(2): 111-114.
16、Yang Y, Li R, Lin D, et al. Automatic identification of myopia based on ocular appearance images using deep learning[ J]. Ann Transl Med, 2020, 8(11): 705.Yang Y, Li R, Lin D, et al. Automatic identification of myopia based on ocular appearance images using deep learning[ J]. Ann Transl Med, 2020, 8(11): 705.
17、Yamada T, Hatt SR, Leske DA, et al. A new computer-based pediatric vision-screening test[ J]. J AAPOS, 2015, 19(2): 157-162.Yamada T, Hatt SR, Leske DA, et al. A new computer-based pediatric vision-screening test[ J]. J AAPOS, 2015, 19(2): 157-162.
18、Rosenfield M, Ciuffreda KJ. Evaluation of the SVOne handheld autorefractor in a pediatric population[ J]. Optom Vis Sci, 2017, 94(2): 159-165.Rosenfield M, Ciuffreda KJ. Evaluation of the SVOne handheld autorefractor in a pediatric population[ J]. Optom Vis Sci, 2017, 94(2): 159-165.
19、Lin H, Long E, Ding X, et al. Prediction of myopia development among Chinese school-aged children using refraction data from electronic medical records: A retrospective, multicentre machine learning study[ J]. PLoS Med, 2018, 15(11): e1002674.Lin H, Long E, Ding X, et al. Prediction of myopia development among Chinese school-aged children using refraction data from electronic medical records: A retrospective, multicentre machine learning study[ J]. PLoS Med, 2018, 15(11): e1002674.
20、Yang X, Chen G, Qian Y, et al. Prediction of myopia in adolescents through machine learning methods[ J]. Int J Environ Res Public Health, 2020, 17(2): 463.Yang X, Chen G, Qian Y, et al. Prediction of myopia in adolescents through machine learning methods[ J]. Int J Environ Res Public Health, 2020, 17(2): 463.
21、唐涛, 范玉琢, 徐琼, 等. 机器学习对青少年近视眼轴增长与近 视度数增加关联性的预测作用[ J]. 中华实验眼科杂志, 2020, 38(2): 134-139.
TANG Tao, FAN Yuzhuo, XU Qiong, et al. A study of the predictive effects of machine learning for the relationship between axial length elongation and the progression of myopia in school-aged children[ J]. Chinese Journal of Experimental Ophthalmology, 2020, 38(2): 134-139.
唐涛, 范玉琢, 徐琼, 等. 机器学习对青少年近视眼轴增长与近 视度数增加关联性的预测作用[ J]. 中华实验眼科杂志, 2020, 38(2): 134-139.
TANG Tao, FAN Yuzhuo, XU Qiong, et al. A study of the predictive effects of machine learning for the relationship between axial length elongation and the progression of myopia in school-aged children[ J]. Chinese Journal of Experimental Ophthalmology, 2020, 38(2): 134-139.
22、Varadarajan AV, Poplin R, Blumer K, et al. Deep learning for predicting refractive error from retinal fundus images[ J]. Invest Ophthalmol Vis Sci, 2018, 59(7): 2861-2868.Varadarajan AV, Poplin R, Blumer K, et al. Deep learning for predicting refractive error from retinal fundus images[ J]. Invest Ophthalmol Vis Sci, 2018, 59(7): 2861-2868.
23、温龙波, 蓝卫忠, 李晓柠, 等. 客观监测近视眼相关环境因素的新 设备“云夹”的准确性和稳定性研究[ J]. 中华眼视光学与视觉科 学杂志, 2017, 19(4): 198-203.
WEN Longbo, LAN Weizhong, LI Xiaoning, et al. Accuracy and stability of ClouclipTM, a novel device to record myopic environmental risks[ J]. Chinese Journal of Optometry Ophthalmology and Visual Science, 2017, 19(4): 198-203.
温龙波, 蓝卫忠, 李晓柠, 等. 客观监测近视眼相关环境因素的新 设备“云夹”的准确性和稳定性研究[ J]. 中华眼视光学与视觉科 学杂志, 2017, 19(4): 198-203.
WEN Longbo, LAN Weizhong, LI Xiaoning, et al. Accuracy and stability of ClouclipTM, a novel device to record myopic environmental risks[ J]. Chinese Journal of Optometry Ophthalmology and Visual Science, 2017, 19(4): 198-203.
24、Cao Y, Lan W, Wen L, et al. An effectiveness study of a wearable device (Clouclip) intervention in unhealthy visual behaviors among schoolage children: A pilot study[ J]. Medicine (Baltimore), 2020, 99(2): e17992.Cao Y, Lan W, Wen L, et al. An effectiveness study of a wearable device (Clouclip) intervention in unhealthy visual behaviors among schoolage children: A pilot study[ J]. Medicine (Baltimore), 2020, 99(2): e17992.
25、居玲, 张常春, 韩香香, 等. 一种基于人工智能的防控近视眼应用 软件[ J]. 中国科技信息, 2020(13): 83-84.
JU Ling, ZHANG Changchun, HAN Xiangxiang, et al. An application software for prevention and control of myopia based on artificial intelligence[ J]. China Science and Technology Information, 2020(13): 83-84.
居玲, 张常春, 韩香香, 等. 一种基于人工智能的防控近视眼应用 软件[ J]. 中国科技信息, 2020(13): 83-84.
JU Ling, ZHANG Changchun, HAN Xiangxiang, et al. An application software for prevention and control of myopia based on artificial intelligence[ J]. China Science and Technology Information, 2020(13): 83-84.
26、周云帆, 蒋沁. 基于移动云计算的儿童和青少年近视防控大数 据平台实践[ J]. 中国数字医学, 2020, 15(6): 123-124, 139.
ZHOU Yunfan, JIANG Qin. Practice of the big data platform for myopia prevention and control of children and adolescents based on mobile cloud computing[ J]. China Digital Medicine, 2020, 15(6): 123-124, 139.
周云帆, 蒋沁. 基于移动云计算的儿童和青少年近视防控大数 据平台实践[ J]. 中国数字医学, 2020, 15(6): 123-124, 139.
ZHOU Yunfan, JIANG Qin. Practice of the big data platform for myopia prevention and control of children and adolescents based on mobile cloud computing[ J]. China Digital Medicine, 2020, 15(6): 123-124, 139.
27、Ohno-Matsui K, Kawasaki R, Jonas JB, et al. International photographic classification and grading system for myopic maculopathy[ J]. Am J Ophthalmol, 2015, 159(5): 877-883.e7.Ohno-Matsui K, Kawasaki R, Jonas JB, et al. International photographic classification and grading system for myopic maculopathy[ J]. Am J Ophthalmol, 2015, 159(5): 877-883.e7.
28、Ikuno Y. Overview of the complications of high myopia[ J]. Retina, 2017, 37(12): 2347-2351.Ikuno Y. Overview of the complications of high myopia[ J]. Retina, 2017, 37(12): 2347-2351.
29、Ohno-Matsui K . W hat is the fundamental nature of pathologic myopia?[ J]. Retina, 2017, 37(6): 1043-1048.Ohno-Matsui K . W hat is the fundamental nature of pathologic myopia?[ J]. Retina, 2017, 37(6): 1043-1048.
30、Zapata MA, Royo-Fibla D, Font O, et al. Artificial intelligence to identify retinal fundus images, quality validation, laterality evaluation, macular degeneration, and suspected glaucoma[ J]. Clin Ophthalmol, 2020, 14: 419-429.Zapata MA, Royo-Fibla D, Font O, et al. Artificial intelligence to identify retinal fundus images, quality validation, laterality evaluation, macular degeneration, and suspected glaucoma[ J]. Clin Ophthalmol, 2020, 14: 419-429.
31、Li Y, Feng W, Zhao X, et al. Development and validation of a deep learning system to screen vision-threatening conditions in high myopia using optical coherence tomography images[ J/OL]. Br J Ophthalmol, 2020, Epub ahead of print.Li Y, Feng W, Zhao X, et al. Development and validation of a deep learning system to screen vision-threatening conditions in high myopia using optical coherence tomography images[ J/OL]. Br J Ophthalmol, 2020, Epub ahead of print.
32、Hemelings R , Elen B, Blaschko MB, et al. Pathological myopia classification with simultaneous lesion segmentation using deep learning[ J]. Comput Methods Programs Biomed, 2021, 199: 105920.Hemelings R , Elen B, Blaschko MB, et al. Pathological myopia classification with simultaneous lesion segmentation using deep learning[ J]. Comput Methods Programs Biomed, 2021, 199: 105920.
33、许迅, 余奇. 人工智能在病理性近视眼底资料解析中的应用和 展望[ J]. 中华眼底病杂志, 2019, 35(5): 427-431.
XU Xun, YU Qi. Application and prospect of artificial intelligence in the analysis of fundus images of pathological myopia[ J]. Chinese Journal of Ocular Fundus Diseases, 2019, 35(5): 427-431.
许迅, 余奇. 人工智能在病理性近视眼底资料解析中的应用和 展望[ J]. 中华眼底病杂志, 2019, 35(5): 427-431.
XU Xun, YU Qi. Application and prospect of artificial intelligence in the analysis of fundus images of pathological myopia[ J]. Chinese Journal of Ocular Fundus Diseases, 2019, 35(5): 427-431.
34、Fu H, Cheng J, Xu Y, et al. Joint optic disc and cup segmentation based on multi-label deep network and polar transformation[ J]. IEEE Trans Med Imaging, 2018, 37(7): 1597-1605.Fu H, Cheng J, Xu Y, et al. Joint optic disc and cup segmentation based on multi-label deep network and polar transformation[ J]. IEEE Trans Med Imaging, 2018, 37(7): 1597-1605.
35、Jiang Y, Duan L, Cheng J, et al. JointRCNN: a region-based convolutional neural network for optic disc and cup segmentation[ J]. IEEE Trans Biomed Eng, 2020, 67(2): 335-343.Jiang Y, Duan L, Cheng J, et al. JointRCNN: a region-based convolutional neural network for optic disc and cup segmentation[ J]. IEEE Trans Biomed Eng, 2020, 67(2): 335-343.
36、Dodo BI, Li Y, Eltayef K, et al. Automatic annotation of retinal layers in optical coherence tomography images[ J]. J Med Syst, 2019, 43(12): 336.Dodo BI, Li Y, Eltayef K, et al. Automatic annotation of retinal layers in optical coherence tomography images[ J]. J Med Syst, 2019, 43(12): 336.
37、Wu Y, Xia Y, Song Y, et al. NFN+: A novel network followed network for retinal vessel segmentation[ J]. Neural Netw, 2020, 126: 153-162.Wu Y, Xia Y, Song Y, et al. NFN+: A novel network followed network for retinal vessel segmentation[ J]. Neural Netw, 2020, 126: 153-162.
38、Tan TE, Anees A, Chen C, et al. Retinal photograph-based deep learning algorithms for myopia and a blockchain platform to facilitate artificial intelligence medical research: a retrospective multicohort study[ J]. Lancet Digit Health, 2021, 3(5): e317-e329.Tan TE, Anees A, Chen C, et al. Retinal photograph-based deep learning algorithms for myopia and a blockchain platform to facilitate artificial intelligence medical research: a retrospective multicohort study[ J]. Lancet Digit Health, 2021, 3(5): e317-e329.
39、Sogawa T, Tabuchi H, Nagasato D, et al. Accuracy of a deep convolutional neural network in the detection of myopic macular diseases using swept-source optical coherence tomography[ J]. PLoS One, 2020, 15(4): e0227240.Sogawa T, Tabuchi H, Nagasato D, et al. Accuracy of a deep convolutional neural network in the detection of myopic macular diseases using swept-source optical coherence tomography[ J]. PLoS One, 2020, 15(4): e0227240.
40、Sánchez-González JM, De-Hita-Cantalejo C, Baustita-Llamas MJ, et al. The combined effect of low-dose atropine with orthokeratology in pediatric myopia control: review of the current treatment status for myopia[ J]. J Clin Med, 202, 9(8): 2371.Sánchez-González JM, De-Hita-Cantalejo C, Baustita-Llamas MJ, et al. The combined effect of low-dose atropine with orthokeratology in pediatric myopia control: review of the current treatment status for myopia[ J]. J Clin Med, 202, 9(8): 2371.
41、Negishi K, Toda I, Ayaki M, et al. Subjective happiness and satisfaction in postoperative anisometropic patients after refractive surgery for myopia[ J]. J Clin Med, 2020, 9(11): 3473.Negishi K, Toda I, Ayaki M, et al. Subjective happiness and satisfaction in postoperative anisometropic patients after refractive surgery for myopia[ J]. J Clin Med, 2020, 9(11): 3473.
42、Bae JI, Yu DS, Kim SY, et al. Effect of optical correction by fully corrected glasses on postural stability[ J]. PLoS One, 2020, 15(7): e0235919.Bae JI, Yu DS, Kim SY, et al. Effect of optical correction by fully corrected glasses on postural stability[ J]. PLoS One, 2020, 15(7): e0235919.
43、Fan Y, Yu Z, Peng Z, et al. Machine learning based strategy surpasses the traditional method for selecting the first trial Lens parameters for corneal refractive therapy in Chinese adolescents with myopia[ J]. Cont Lens Anterior Eye, 2021, 44(3): 101330.Fan Y, Yu Z, Peng Z, et al. Machine learning based strategy surpasses the traditional method for selecting the first trial Lens parameters for corneal refractive therapy in Chinese adolescents with myopia[ J]. Cont Lens Anterior Eye, 2021, 44(3): 101330.
44、Li SM, Kang MT, Wang NL, et al. Wavefront excimer laser refractive surgery for adults with refractive errors[ J]. Cochrane Database Syst Rev, 2020, 12: CD012687.Li SM, Kang MT, Wang NL, et al. Wavefront excimer laser refractive surgery for adults with refractive errors[ J]. Cochrane Database Syst Rev, 2020, 12: CD012687.
45、Redd TK, Campbell JP, Chiang MF, et al. Artificial intelligence for refractive surgery screening: finding the balance between myopia and hype-ropia[ J]. JAMA Ophthalmol, 2020, 138(5): 526-527.Redd TK, Campbell JP, Chiang MF, et al. Artificial intelligence for refractive surgery screening: finding the balance between myopia and hype-ropia[ J]. JAMA Ophthalmol, 2020, 138(5): 526-527.
46、Castro-Luna G, Pérez-Rueda A. A predictive model for early diagnosis of keratoconus[ J]. BMC Ophthalmol, 2020, 20(1): 263.Castro-Luna G, Pérez-Rueda A. A predictive model for early diagnosis of keratoconus[ J]. BMC Ophthalmol, 2020, 20(1): 263.
47、Lavric A, Valentin P. KeratoDetect: keratoconus detection algorithm using convolutional neural networks[ J]. Comput Intell Neurosci, 2019, 2019: 8162567.Lavric A, Valentin P. KeratoDetect: keratoconus detection algorithm using convolutional neural networks[ J]. Comput Intell Neurosci, 2019, 2019: 8162567.
48、Xie Y, Zhao L, Yang X, et al. Screening candidates for refractive surgery with corneal tomographic-based deep learning[ J]. JAMA Ophthalmol, 2020, 138(5): 519-526.Xie Y, Zhao L, Yang X, et al. Screening candidates for refractive surgery with corneal tomographic-based deep learning[ J]. JAMA Ophthalmol, 2020, 138(5): 519-526.
49、Cui T, Wang Y, Ji S, et al. Applying machine learning techniques in nomogram prediction and analysis for SMILE treatment[ J]. Am J Ophthalmol, 2020, 210: 71-77.Cui T, Wang Y, Ji S, et al. Applying machine learning techniques in nomogram prediction and analysis for SMILE treatment[ J]. Am J Ophthalmol, 2020, 210: 71-77.
50、Achiron A, Gur Z, Aviv U, et al. Predicting refractive surgery outcome: machine learning approach with big data[ J]. J Refract Surg, 2017, 33(9): 592-597.Achiron A, Gur Z, Aviv U, et al. Predicting refractive surgery outcome: machine learning approach with big data[ J]. J Refract Surg, 2017, 33(9): 592-597.
51、Celi LA, Citi L, Ghassemi M, et al. The PLoS One collection on machine learning in health and biomedicine: Towards open code and open data[ J]. PLoS One, 2019, 14(1): e0210232.Celi LA, Citi L, Ghassemi M, et al. The PLoS One collection on machine learning in health and biomedicine: Towards open code and open data[ J]. PLoS One, 2019, 14(1): e0210232.
52、Wang S, Yu L, Yang X, et al. Patch-based output space adversarial learning for joint optic disc and cup segmentation[ J]. IEEE Trans Med Imaging, 2019, 38(11): 2485-2495.Wang S, Yu L, Yang X, et al. Patch-based output space adversarial learning for joint optic disc and cup segmentation[ J]. IEEE Trans Med Imaging, 2019, 38(11): 2485-2495.
53、Zhang W, Zhao X, Chen Y, et al. DeepUWF: An automated ultra-widefield fundus screening system via deep learning[ J]. IEEE J Biomed Health Inform, 2021, 25(8): 2988-2996.Zhang W, Zhao X, Chen Y, et al. DeepUWF: An automated ultra-widefield fundus screening system via deep learning[ J]. IEEE J Biomed Health Inform, 2021, 25(8): 2988-2996.
1、陆兵. 面向彩色眼底图像的眼科常见病智能辅助诊断模型研究[D].湖州师范学院,2022.Lu B. Research on intelligent aided diagnosis model of common ophthalmic diseases for color fundus images[D]. Huzhou Teachers College: 2022.
2、李铁铮.青少年近视防控研究综合谈[J].中国眼镜科技杂志,2022(10):111-115.Li TZ. A comprehensive discussion on the prevention and control of juvenile myopia[J]. China Glasses Sci Technol Mag, 2022(10): 111-115.
1、江苏省自然科学基金 (BK20191233)。This work was supported by the Natural Science Foundation of Jiangsu Province, China (BK20191233)()
上一篇
下一篇
其他期刊
  • 眼科学报

    主管:中华人民共和国教育部
    主办:中山大学
    承办:中山大学中山眼科中心
    主编:林浩添
    主管:中华人民共和国教育部
    主办:中山大学
    浏览
  • Eye Science

    主管:中华人民共和国教育部
    主办:中山大学
    承办:中山大学中山眼科中心
    主编:林浩添
    主管:中华人民共和国教育部
    主办:中山大学
    浏览
推荐阅读
出版者信息
中山眼科



中山大学
目录