• Users Online: 26483
  • Home
  • Print this page
  • Email this page

   Table of Contents      
COMMENTARY
Year : 2020  |  Volume : 68  |  Issue : 7  |  Page : 1347-1348

Commentary: Artificial intelligence in ophthalmology: Potential challenges and way ahead


Dr. Rajendra Prasad Centre for Ophthalmic Sciences, All India Institute of Medical Sciences, Ansari Nagar, New Delhi, India

Date of Web Publication25-Jun-2020

Correspondence Address:
Dr. Atul Kumar
Retina Services, Dr. Rajendra Prasad Centre for Ophthalmic Sciences, All India Institute of Medical Sciences, Ansari Nagar, New Delhi - 110 029
India
Login to access the Email id

Source of Support: None, Conflict of Interest: None


DOI: 10.4103/ijo.IJO_737_20

Rights and Permissions

How to cite this article:
Agarwal D, Kumar A. Commentary: Artificial intelligence in ophthalmology: Potential challenges and way ahead. Indian J Ophthalmol 2020;68:1347-8

How to cite this URL:
Agarwal D, Kumar A. Commentary: Artificial intelligence in ophthalmology: Potential challenges and way ahead. Indian J Ophthalmol [serial online] 2020 [cited 2024 Mar 29];68:1347-8. Available from: https://journals.lww.com/ijo/pages/default.aspx/text.asp?2020/68/7/1347/287566



Artificial intelligence (AI) has made tremendous advancements in the field of health care. Newer applications of AI are making a fruitful impact in ophthalmology as well. They have shown potential in diseases like diabetic retinopathy (DR), retinopathy of prematurity, age-related macular degeneration (ARMD), glaucoma, cataract grading, refractive error prediction, etc.[1],[2] The present study has described in detail the uses of AI and its role in ophthalmology.[3] Currently, the role of AI is mainly in medical image identification and auxiliary diagnosis. AI has helped in achieving quality, accessibility, and availability in the field of health care. With the limited number of specialists and increasing population, the limitation of resources is always there. This can lead to late diagnosis, late initiation of treatment, and late referral of common treatable and preventable causes of blindness. Machine learning is one of the most commonly used principles in AI.[4] This involves the training of machine by providing a large amount of data followed by validation. The machine uses various models to draw its own inference pattern like convoluted neural networks. Keeping in mind its enormous potential, the NITI Aayog has formulated National Strategy for AI in June 2018 and is mandated to establish a national program on AI in critical sectors like agriculture, health care, etc.[5]

There are also some unanswered questions and challenges which must be addressed in the future. The legal implications of AI in India and other parts of the world are one of the major hurdles. The legal issues involved with the development and implementation of AI algorithms are considerable. Regulation regarding legal causes of action such as medical malpractice and product liability, intellectual property rights, data security, and patient privacy are major contentious issues.[5] It is very difficult to set responsibility of the decisions/actions made by the machine on the developer or the user/customer, in case of misdiagnosis. The AI is not infallible. Current AI applications cater to diagnose only a single disease. A negative (normal report) finding of AI may give a false sense of security to the primary physician and patient leading to delay in the management of other serious eye conditions. For example, IDX DR can only diagnose and refer cases of DR. It will miss other co-existent ocular conditions like glaucoma, cataract, dry eye, etc., Thus, AI cannot replace a comprehensive dilated ocular examination by an ophthalmologist. The machine can have high false-positive rates which can lead to unnecessary fear and overtreatment in patients. For example, in ARMD, the machine can pick up the smallest of drusens which may not warrant any treatment. Most of the AI-based retinal applications examine patients in undilated pupils. These can miss important findings in the peripheral retina which could be early markers of disease. Eye diseases can have regional variations as well. AI needs to be customized to the region and target population. The US-based data set cannot be strictly applied to Indian eyes. One such welcome step is Google Inc. collaborating with Aravind Eye Hospital to develop an AI algorithm for DR screening in India. Patient privacy and confidentiality of highly sensitive patient information, also need some legal regulation.

As it is still in the evolving stage, the high cost of AI can be a hindrance in resource-limited countries. It can only be used for diseases that have a high prevalence and high morbidity. Also, we don't know how the machine draws a certain pattern.[4] A good quality image data set for training and experienced annotators for validation are the basis of a good AI-based algorithm. Nguyen et al. have described how a machine can make a mistake.[6] It can also give wrong results if the relationship between input and expected output variables is complex as the machine is unable to build the model.[7]

In conclusion, AI can help in providing relief to the overburdened healthcare system and also help in reducing social inequality. The downsides of AI should always be kept in mind. AI cannot replace doctors but it can make them better.



 
  References Top

1.
Padhy SK, Takkar B, Chawla R, Kumar A. Artificial intelligence in diabetic retinopathy: A natural step to the future. Indian J Ophthalmol 2019;67:1004.  Back to cited text no. 1
[PUBMED]  [Full text]  
2.
Akkara JD, Kuriakose A. Role of artificial intelligence and machine learning in ophthalmology. Kerala J Ophthalmol 2019;31:150-60.  Back to cited text no. 2
  [Full text]  
3.
Dutt S, Sivaraman A, Savoy F, Rajalakshmi R. Insights into the growing popularity of artificial intelligence in ophthalmology. Indian J Ophthalmol 2020;68:1339-46.  Back to cited text no. 3
  [Full text]  
4.
Coyner AS, Campbell JP, Chiang MF. Demystifying the jargon: The bridge between ophthalmology and artificial intelligence. Ophthalmol Retina 2019;3:291-3.  Back to cited text no. 4
    
5.
National Strategy on Artificial Intelligence. NITI Aayog. Available from: https://niti.gov.in/national-strategy-artificial-intelligence. [Last accessed on 2020 Mar 29].  Back to cited text no. 5
    
6.
Nguyen A, Yosinski J, Clune J. Deep neural networks are easily fooled: high confidence predictions for unrecognizable images. In: Proceedings of the IEEE conference on computer vision and pattern recognition. 2015. p. 427-36.  Back to cited text no. 6
    
7.
Du XL, Li WB, Hu BJ. Application of artificial intelligence in ophthalmology. Int J Ophthalmol 2018;11:1555-61.  Back to cited text no. 7
    




 

Top
 
 
  Search
 
    Similar in PUBMED
   Search Pubmed for
   Search in Google Scholar for
    Access Statistics
    Email Alert *
    Add to My List *
* Registration required (free)  

 
  In this article
References

 Article Access Statistics
    Viewed1389    
    Printed81    
    Emailed0    
    PDF Downloaded182    
    Comments [Add]    

Recommend this journal