您的位置：山东大学 -> 科技期刊社 -> 《山东大学学报（医学版）》

山东大学学报 (医学版) ›› 2020, Vol. 58 ›› Issue (11): 17-23.doi: 10.6040/j.issn.1671-7554.0.2020.1227

• 眼科人工智能新进展专题 • 上一篇    下一篇

眼科人工智能的算法新进展

戈宗元*(),贺婉佶,琚烈,姚轩,王璘,黄烨霖,杨志文,熊健皓,包怡宁,李明,张兵,赵昕   

  1. 澳大利亚莫纳什大学工程学院, 维多利亚州 墨尔本 3800
  • 收稿日期:2020-08-29 出版日期:2020-11-10 发布日期:2020-11-04
  • 通讯作者: 戈宗元 E-mail:zongyuan.ge@monash.edu
  • 作者简介:戈宗元，博士，博士研究生导师，现任澳大利亚莫纳什大学(Monash University)工程学院高级讲师/研究员，澳大利亚Nvidia人工智能技术中心深度学习专家，Airdoc澳大利亚研究中心首席科学家。参与主持科研项目800万澳元，发表SCI文章50余篇，获得8项专利，参与编写专著1部。其对医学人工智能的开发和研究有着浓厚的兴趣，在统计分析、机器学习、计算机视觉和医疗人工智能研究方面有很强的专业知识背景，已主持并参与了6项国际研究项目。其所涉及的专业领域涵盖了皮肤病学、眼科学和放射学，合作对象包括了工业界研究巨头如IBM Watson健康和医疗服务提供商Molemap。其研究成果已成功转型为健康产品提供给诸如Specsavers与Bupa等客户。2017年被海德堡获奖者论坛基金会的科学委员会评选为计算机与数学领域200名杰出青年研究者之一，2017年获IBM科学研究成就奖和IBM经理人票选奖，2019年获得莫纳什卓越成就奖

New developments in ophthalmic AI algorithms

Zongyuan GE*(),Wanji HE,Lie JU,Xuan YAO,Lin WANG,Yelin HUANG,Zhiwen YANG,Jianhao XIONG,Yining BAO,Ming LI,Bing ZHANG,Xin ZHAO   

  1. Monash e-Research Center and Faculty of Engineering, Monash University, Melbourne 3800, Victoria, Australia
  • Received:2020-08-29 Online:2020-11-10 Published:2020-11-04
  • Contact: Zongyuan GE E-mail:zongyuan.ge@monash.edu

RichHTML

52

PDF (PC)

416

摘要：

随着高效的深度神经网络算法、大量高质量医学数据、低成本大规模计算机并行设备的普及，近年来人工智能在眼科领域、院内眼科疾病筛查和院外体检中心都取得了大规模的应用。对某些特定疾病如眼底糖网已经达到甚至超过了大多数全科大夫的水准。在本文中我们以分类、检测、分割、域适应等基础算法为引子，梳理、分析出人工智能在眼科应用中的优势和不足，以便更好地构想未来的研究方向。

关键词: 深度神经网络, 眼科疾病检测, 人工智能, 计算机视觉

Abstract:

Deep neural networks, combined with high quality annotated medical data and low-cost GPU devices, have been successfully implemented in the field of ophthalmology. Impressive grounding outcomes have occurred in both in-hospital and out-hospital scenarios. Some of the published results demonstrated AI could achieve better diagnosis performance than general practice in the diagnosis of certain retinal diseases. In this article, we will discuss how classification, detection, segmentation and domain adaptation play their roles in AI ophthalmology. We will also discuss the limitations of the current state-of-the-art (SOTA) algorithms, hoping to provide prevision for future research in this field.

Key words: Deep neural networks, Retinal disease diagnosis, Artificial intelligence, Computer vision

中图分类号: 

  • R770.4
1 路寻. 卓越的人工智能科学家——马文·明斯基[J]. 自然辩证法通讯, 2010, 32 (2): 104- 111, 128.
2 Govindaiah A, Smith RT, Bhuiyan A. A new and improved method for automated screening of age-related macular degeneration using ensemble deep neural networks[C]. Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), 2018: 702-705.
3 Russakoff DB , Lamin A , Oakley JD , et al. Deep learning for prediction of amd progression: A pilot study[J]. Invest Ophthalmol Vis Sci, 2019, 60 (2): 712- 722.
doi: 10.1167/iovs.18-25325
4 Ting DSW , Cheung CY , Lim G , et al. Development and validation of a deep learning system for diabetic retinopathy and related eye diseases using retinal images from multiethnic populations with diabetes[J]. JAMA, 2017, 318 (22): 2211- 2223.
doi: 10.1001/jama.2017.18152
5 Brown JM , Campbell JP , Beers A , et al. Automated diagnosis of plus disease in retinopathy of prematurity using deep convolutional neural networks[J]. JAMA Ophthalmol, 2018, 136 (7): 803- 810.
doi: 10.1001/jamaophthalmol.2018.1934
6 Kesim C , Tas AY , Karslıoglu MZ , et al. Validation results of a deep learning algorithm for detection of diabetic retinopathy with lesion localization from retinal fundus photographs[J]. Invest Ophthalmol Vis Sci, 2020, 62 (7): 1626.
7 Lakra A, Tripathi P, Keshari R, et al. Segdensenet: Iris segmentation for pre and post cataract surgery[J/OL]. arXiv, 2018: 1801.10100[2020-08-16]. https://ui.adsabs.harvard.edu/abs/2018arXiv180110100L.
8 Dong Y, Zhang Q, Qiao Z, et al. Classification of cataract fundus image based on deep learning[C]. Proceedings of IEEE International Conference on Imaging Systems and Techniques, 2017: 1-5.
9 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.
doi: 10.1167/iovs.18-23887
10 Yau JW , Rogers SL , Kawasaki R , et al. Global prevalence and major risk factors of diabetic retinopathy[J]. Diabetes Care, 2012, 35 (3): 556- 564.
11 Ortiz A , Munilla J , Gorriz JM , et al. Ensembles of deep learning architectures for the early diagnosis of the alzheimer's disease[J]. Int J Neural Syst, 2016, 26 (7): 1650025.
doi: 10.1142/S0129065716500258
12 Gulshan V , Peng L , Coram M , et al. Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs[J]. JAMA, 2016, 316 (22): 2402- 2410.
doi: 10.1001/jama.2016.17216
13 Diaz-Pinto A , Morales S , Naranjo V , et al. Cnns for automatic glaucoma assessment using fundus images: An extensive validation[J]. Biomed Eng Online, 2019, 18 (1): 29.
doi: 10.1186/s12938-019-0649-y
14 Wang X, Ju L, Zhao X, et al. Retinal abnormalities recognition using regional multitask learning[C]. International Conference on Medical Image Computing and Computer-Assisted Intervention, 2019: 30-38.
15 Ruder S. An overview of multi-task learning in deep neural networks[J/OL]. arXiv, 2017: 1706.05098[2020-08-16]. https://ui.adsabs.harvard.edu/abs/2017arXiv170605098R.
16 Ravindran S . How artificial intelligence is helping to prevent blindness[J]. Nature, 2019, 568 (7751): 6.
17 Nair A , Merkel S , Straub J , et al. Real-time pupil detection for fundus imager using single-shot detector and hard- negative training[J]. Invest Ophthalmol Vis Sci, 2020, 61 (9): PB00110- PB00110.
18 Mitra A , Banerjee PS , Roy S , et al. The region of interest localization for glaucoma analysis from retinal fundus image using deep learning[J]. Comput Methods Programs Biomed, 2018, 165 (2018): 25- 35.
19 Selvaraju RR, Das A, Vedantam R, et al. Grad-cam: Why did you say that?[J/OL]. arXiv, 2016: 1611.07450.[2020-08-16]. https://ui.adsabs.harvard.edu/abs/2016arXiv161107450S.
20 Haloi M. Improved microaneurysm detection using deep neural networks[J/OL]. arXiv, 2015: 1505.04424[2020-08-16]. https://ui.adsabs.harvard.edu/abs/2015arXiv-150504424H.
21 Khojasteh P , Passos Junior LA , Carvalho T , et al. Exudate detection in fundus images using deeply-learnable features[J]. Comput Biol Med, 2019, 104 (2019): 62- 69.
22 Mitani A , Huang A , Venugopalan S , et al. Detection of anaemia from retinal fundus images via deep learning[J]. Nat Biomed Eng, 2020, 4 (1): 18- 27.
23 Quellec G , Lamard M , Conze PH , et al. Automatic detection of multiple pathologies in fundus photographs[J]. Invest Ophthalmol Vis Sci, 2020, 61 (7): 1641- 6141.
24 Fang L , Cunefare D , Wang C , et al. Automatic segmentation of nine retinal layer boundaries in oct images of non-exudative amd patients using deep learning and graph search[J]. Biomed Opt Express, 2017, 8 (5): 2732- 2744.
doi: 10.1364/BOE.8.002732
25 De Fauw J , Ledsam JR , Romera-Paredes B , et al. Clinically applicable deep learning for diagnosis and referral in retinal disease[J]. Nat Med, 2018, 24 (9): 1342- 1350.
doi: 10.1038/s41591-018-0107-6
26 Ma W, Yu S, Ma K, et al. Multi-task neural networks with spatial activation for retinal vessel segmentation and artery/vein classification[C]. Proceedings of the Medical Image Computing and Computer Assisted Intervention-MICCAI 2019, Cham, F, 2019: 769-778.
27 Sevastopolsky A . Optic disc and cup segmentation methods for glaucoma detection with modification of u-net convolutional neural network[J]. Pattern Recog& Ima Anal, 2017, 27 (3): 618- 624.
28 Sarhan MH, Albarqouni S, Yigitsoy M, et al. Multi-scale microaneurysms segmentation using embedding triplet loss[C]. Proceedings of the Medical Image Computing and Computer Assisted Intervention-MICCAI 2019, Cham, F, 2019: 174-182.
29 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.
doi: 10.1109/TMI.2018.2791488
30 Shelhamer E , Long J , Darrell T . Fully convolutional networks for semantic segmentation[J]. IEEE Trans Pattern Anal Mach Intell, 2017, 39 (4): 640- 651.
doi: 10.1109/TPAMI.2016.2572683
31 Ronneberger O, Fischer P, Brox T. U-net: Convolutional networks for biomedical image segmentation[J/OL]. arXiv, 2015: 1505.04597[2020-08-16]. https://ui.adsabs.harvard.edu/abs/2015arXiv150504597R.
32 Chen LC, Zhu Y, Papandreou G, et al. Encoder-decoder with atrous separable convolution for semantic image segmentation[J/OL]. arXiv, 2018: 1802.02611[2020-08-16]. https://ui.adsabs.harvard.edu/abs/2018arXiv180202611C.
33 Yan X, Jiang W, Shi Y, et al. Ms-nas: Multi-scale neural architecture search for medical image segmentation[J/OL]. arXiv, 2020: 2007.06151[2020-08-16]. https://ui.adsabs.harvard.edu/abs/2020arXiv200706-151Y.
34 Jiang Y, Tan N. Retinal vessel segmentation based on conditional deep convolutional generative adversarial networks[J/OL]. arXiv, 2018: 1805.04224[2020-08-16]. https://ui.adsabs.harvard.edu/abs/2018arXiv180504-224J.
35 Redko I, Habrard A, Morvant E, et al. 2 - domain adaptation problem[M]// Redko I, Habrard A, Morvant E, et al. Advances in domain adaptation theory. Amsterdam: Elsevier, 2019: 21-36.
36 Ma Y, Lao S, Takikawa E, et al. Discriminant analysis in correlation similarity measure space[C]. Proceedings of the 24th international conference on Machine learning, 2007: 577-584.
37 Bruzzone L , Marconcini M . Domain adaptation problems: A dasvm classification technique and a circular validation strategy[J]. IEEE Trans Pattern Anal Mach Intell, 2010, 32 (5): 770- 787.
doi: 10.1109/TPAMI.2009.57
38 Gheisari M , Baghshah MS . Unsupervised domain adaptation via representation learning and adaptive classifier learning[J]. Neurocomputing, 2015, 165 (OCT.1): 300- 311.
39 Hinton G, Vinyals O, Dean J. Distilling the knowledge in a neural network[J/OL]. arXiv, 2015: 1503.02531[2020-08-16]. https://ui.adsabs.harvard.edu/abs/2015arXiv150302531H.
40 Borgwardt KM , Gretton A , Rasch MJ , et al. Integrating structured biological data by kernel maximum mean discrepancy[J]. Bioinformatics, 2006, 22 (14): E49- E57.
doi: 10.1093/bioinformatics/btl242
41 Ghifary M, Bastiaan Kleijn W, Zhang M. Domain adaptive neural networks for object recognition[J/OL]. arXiv, 2014: 1409.6041[2020-08-16]. https://ui.adsabs.harvard.edu/abs/2014ar, Xiv1409.6041G.
42 Li Y , Wang N , Shi J , et al. Adaptive batch normalization for practical domain adaptation[J]. Pattern Recognition, 2018, 80 (2018): 109- 117.
43 Gopalan R, Ruonan L, Chellappa R. Domain adaptation for object recognition: An unsupervised approach[C]. International Conference on Computer, 2011: 999-1006.
44 Liu M-Y, Tuzel O. Coupled generative adversarial networks[J/OL]. arXiv, 2016: 1606.07536[2020-08-16]. https://ui.adsabs.harvard.edu/abs/2016arXiv1606075-36L.
45 Ganin Y, Lempitsky V. Unsupervised domain adaptation by backpropagation[J/OL]. arXiv, 2014: 1409.7495[2020-08-16]. https://ui.adsabs.harvard.edu/abs/2014arXiv1409.7495G.
46 Bousmalis K, Trigeorgis G, Silberman N, et al. Domain separation networks[J/OL]. arXiv, 2016: 1608.06019[2020-08-16]. https://ui.adsabs.harvard.edu/abs/2016arXiv160806019B.
47 Yi Z, Zhang H, Tan P, et al. Dualgan: Unsupervised dual learning for image-to-image translation[J/OL]. arXiv, 2017: 1704.02510[2020-08-16]. https://ui.adsabs.harvard.edu/abs/2017arXiv170402510Y.
48 Zhu J, Park T, Isola P, et al. Unpaired image-to-image translation using cycle-consistent adversarial networks[C].IEEE International Conference on Computer Vision (ICCV), 2017: 2242-2251.
49 Yang Y, Soatto S. Fda: Fourier domain adaptation for semantic segmentation[J/OL]. arXiv, 2020: 2004.05498[2020-08-16]. https://ui.adsabs.harvard.edu/abs/2020arXiv200405498Y.
50 Johnson JM , Khoshgoftaar TM . Survey on deep learning with class imbalance[J]. Journal of Big Data, 2019, 6 (1): 1- 54.
51 Xiong J, He AW, Fu M, et al. Improve unseen domain generalization via enhanced local color transformation[C]. Proceedings of the Medical Image Computing and Computer Assisted Intervention-MICCAI 2020, Cham, F, 2020: 433-443.
52 Davide C . Can we open the black box of ai?[J]. Nature, 2016, 538 (7623): 20.
doi: 10.1038/538020a
53 Ge Z, Demyanov S, Chen Z, et al. Generative openmax for multi-class open set classification[J/OL]. arXiv, 2017: 1707.07418[2020-08-16]. https://ui.adsabs.harvard.edu/abs/2017arXiv170707418G.
54 Heaven WD. Google's medical ai was super accurate in a lab. Real life was a different story[EB/OL]. (2020-08-16)[2020-08-18]. https://www.technologyreview.com/2020/04/27/1000658/google-medical-ai-accurate-lab-real-life-clinic-covid-diabetes-retina-disease/.
55 Airdoc取得nmpa人工智能医疗器械三类注册证[EB/OL]. (2020-08-16)[2020-08-18]. https://www.sohu.com/a/412357707_102972.
[1] 刘琚,吴强,于璐跃,林枫茗. 基于深度学习的脑肿瘤图像分割[J]. 山东大学学报 (医学版), 2020, 1(8): 42-49, 73.
[2] 李新钢,张鑫,陈安静. 当代脑计划研究进展[J]. 山东大学学报 (医学版), 2020, 1(8): 5-9, 21.
[3] 赵继宗. 神经外科学是脑科学研究的一支主力军[J]. 山东大学学报 (医学版), 2020, 1(8): 1-4.
[4] 曲毅,张焕开,宋先,初宝睿. 人工智能诊断系统在视网膜疾病的研究进展[J]. 山东大学学报 (医学版), 2020, 58(11): 39-44.
[5] Carol Y. Cheung, 冉安然. 青光眼影像人工智能深度学习研究现状与展望[J]. 山东大学学报 (医学版), 2020, 58(11): 24-32.
[6] 林浩添,李龙辉,陈睛晶. 儿童眼病的人工智能研究进展[J]. 山东大学学报 (医学版), 2020, 58(11): 11-16.
[7] 何明光,刘驰,李治玺. 人工智能在眼科真实临床场景的应用：机遇和挑战[J]. 山东大学学报 (医学版), 2020, 58(11): 1-10.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
[1] 索东阳,申飞,郭皓,刘力畅,杨惠敏,杨向东. Tim-3在药物性急性肾损伤动物模型中的表达及作用机制[J]. 山东大学学报 (医学版), 2020, 1(7): 1 -6 .
[2] 马青源,蒲沛东,韩飞,王超,朱洲均,王维山,史晨辉. miR-27b-3p调控SMAD1对骨肉瘤细胞增殖、迁移和侵袭作用的影响[J]. 山东大学学报 (医学版), 2020, 1(7): 32 -37 .
[3] 李宁,李娟,谢艳,李培龙,王允山,杜鲁涛,王传新. 长链非编码RNA AL109955.1在80例结直肠癌组织中的表达及对细胞增殖与迁移侵袭的影响[J]. 山东大学学报 (医学版), 2020, 1(7): 38 -46 .
[4] 徐玉香,刘煜东,张蓬,段瑞生. 101例脑小血管病患者脑微出血危险因素的回顾性分析[J]. 山东大学学报 (医学版), 2020, 1(7): 67 -71 .
[5] 罗昕,何兵,聂清生,侯震波,董军,李玉花,曾祥芹,刘伟,孔德民,曹金凤. 磁共振扩散加权成像单指数模型与扩散峰度成像模型在61例肾透明细胞癌分级中的对比[J]. 山东大学学报 (医学版), 2020, 1(7): 89 -95 .
[6] 张迪,于猛,刘霞. 神经调控技术简述[J]. 山东大学学报 (医学版), 2020, 1(8): 50 -60 .
[7] 张伟,谭文浩,李贻斌. 基于深度强化学习的四足机器人运动控制发展现状与展望[J]. 山东大学学报 (医学版), 2020, 1(8): 61 -66 .
[8] 陈安静,张训. 靶向小类泛素化修饰的胶质瘤治疗新策略[J]. 山东大学学报 (医学版), 2020, 1(8): 88 -94 .
[9] 张洪彬,赵寒辉,王素霞,周鹏,贺青卿,王延群,丁伟平,柳刚. 303例甲状旁腺切除术围术期观察及术后严重低钙血症危险因素分析[J]. 山东大学学报 (医学版), 2020, 1(9): 14 -20 .
[10] 孙薏丰,高玉,梁永媛,高杨. CPLX2在30例肝癌组织的表达及其对体外细胞增殖与侵袭的影响[J]. 山东大学学报 (医学版), 2020, 1(9): 34 -39 .
版权所有 © 2018 《山东大学学报(医学版)》编辑部　
地址:山东大学科技期刊社（济南市历城区山大南路27号山东大学中心校区明德楼B座721室） 电话: 0531-88366918 E-mail: xbyxb@sdu.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发