Sophistication in AI/ML application for Diabetic Retinopathy diagnosis has tremendous research scope by Anusha Ashwin, @ashwin_anusha

India – better known as the Diabetes Capital of the World – has been battling with this non-communicable lifestyle associated disease for long. According to the International Diabetes Federation, India’s diabetes population totalled 74 million in 2017 and it is estimated to reach 134.3 million people by 2045.

Diabetes in India is an epidemic that presents itself with several other complications when left uncontrolled or untreated. And notably, complications in the eye and vision are the most commonly associated negative impact of diabetes in India. Among the complications in the eye, the most prevalent type of diabetic eye diseases is Diabetic Retinopathy – which on early diagnosis and intervention is completely curable.

Economic burden of Diabetic Retinopathy 

To explain the impact of Diabetic Retinopathy on Indian population, and how by deploying Artificial Intelligence, the screening and diagnosis has undergone an evolution, Dr Sheila John, Head of Teleophthalmology, Sankara Nethralaya, Chennai, spares her time for HCITExpert Blog

Sankara Nethralaya is one of the pioneering institutions to initiate telemedicine in the country. Much before broadband was made available in India, Sankara Nethralaya is known to have used the satellite internet connection provided by ISRO for setting up of mobile vans that could take eye examination centres to rural parts of India. Dr Sheila John has been instrumental in setting up fully-equipped vans with all ophthalmic and telemedicine facilities that can visit the rural areas where patients are examined by a team of optometrists and social workers .She is further responsible to ensure that teleconsultations are provided by the ophthalmologists at the base hospital located in the metro city. 

In the past few years, she has been working in collaboration with the Healthcare Technology Innovation Centre, Indian Institute of Technology, Madras, to research the application of AI in Diabetic Retinopathy. She also has sustained research interests in organization and incorporation of Electronic Medical Records and Tele–ophthalmology in offsite camps.

Diabetic Retinopathy (DR) is asymptomatic and affects nearly 20% of the 70 million diabetic patients in India. DR generally affects persons with diabetes for many years and is harmful if left untreated, raising the possibility of blindness. With aging as well as with less regulated blood sugar and blood pressure levels, the possibility of causing diabetic retinopathy is considered to rise among Indian demographics.

Being asymptomatic, the diabetic patient hardly realizes the condition in the eye until vision starts to become hazy and weak. Dr Sheila emphasises that it is mandatory for diabetics to undergo an annual eye examination to rule out the possibility of DR. Timely and accessible DR screening is critical for early DR detection and in prevention of needless blindness. 

Now the economic impact of DR in Indian population is also a big matter of concern for the country. Nearly 70% of the population, especially those in rural areas lack access to quality eye care and specialists, given the very low ophthalmologists to population ratio of 1:100000 in some areas. Since, most of the rural population is dominated by the agrarian community, loss of vision owing to DR can directly affect the output of agricultural produce and also the sole breadwinner of the family becomes incapable of supporting the family and in turn becomes a burden on the family and government. The public health burden owing to DR on urban populations also has the similar effect by undiagnosed diabetics, who are prone to retinopathy conditions. 

Telemedicine in Diabetic Retinopathy

Before understanding from Dr Sheila John on why telemedicine is necessary for DR diagnosis, it is important for us to know a little more about Diabetic Retinopathy’s manifestation in a diabetic patient. 

Diabetic retinopathy is an eye condition that occurs when the blood vessels in the back of the eye leak fluid, protein and fat, and bleed as a complication of diabetes. The longer a person is diabetic, higher levels of lipids, poor kidney function, the higher their chance of developing diabetic retinopathy. There can be a partial vision loss to complete vision loss in one or both the eyes. 

In severe cases, when blood sugar levels run too high, small blood vessels in the retina can become damaged, causing them to leak fluid or bleed. The body then regenerates new blood vessels to compensate for the damaged ones, but these new blood vessels may also bleed or leak easily.

At first, diabetic retinopathy may have no symptoms or can cause only mild vision problems. As time goes by, it can potentially lead to vision loss. Since the early stages of this condition may not produce any symptoms, Dr Sheila insists that it is vital for all patients with type 1 or type 2 diabetes to undergo an annual comprehensive dilated eye examination in order to accurately diagnose Diabetic Retinopathy. 

The stages of diabetic retinopathy: non-proliferative (Mild, Moderate and severe), diabetic maculopathy and proliferative retinopathy. In the early, non-proliferative stages, symptoms may be very minimal or non-existent. In sight threatening retinopathy (diabetic maculopathy and Proliferative diabetic retinopathy), there may be noticeable vision changes that can indicate significant deterioration of eye health and vision loss. Sight threatening retinopathy is severe and advanced stages of diabetic retinopathy which requires immediate care and treatment by ophthalmologists 

Dr Sheila says that it is important to catch the condition as early as in the non-proliferative stage where mild symptoms appear. Catching the symptoms at this early stage, especially among rural diabetic patients is necessary and this is where telemedicine plays a significant role in early diagnosis and intervention. 

The biggest advantage in early diagnosis is the cost saving approach for the patient. As the financial capacity of the patient matters, telemedicine makes it possible for remote delivery of the healthcare at an early intervention stage, which would otherwise not be possible for a patient in poor economic conditions to travel to locations that have speciality care.

The telemedicine set-up

As mentioned earlier, Sankara Nethralaya has mobile vans fully equipped with all facilities for eye examinations in remote locations. The vans house state-of-the-art ophthalmic equipment for comprehensive eye examination, fundus cameras that are portable, handheld or smartphone-based devices. The vans have telehealth software-loaded laptops that maintain Electronic Medical Records, and can enable video conferencing to the base hospital. It is even possible for these vans to administer primary treatment medicines to the patient at the campsite. 

As for the internet connectivity the vans carry data cards with 2Mbps internet speed or with available bandwidth of 120 to 150 kbps local internet. Real-time image sharing with annotation from the campsite to the base hospital is made possible for optometrists to mediate consultations with senior ophthalmologist at the base hospital in these vans. 

Artificial Intelligence in DR diagnosis

Artificial Intelligence-based deep learning algorithm technology offers the potential to increase the efficiency and accessibility of DR screening programs. Dr Sheila John says that the feasibility of using a smartphone-based retinal fundus camera with in-built automated Artificial Intelligence (AI) algorithm is a boon for the eye healthcare givers as well as for the patients. 

The algorithms are built to identify DR in specific, in diabetic patients, who undergo a primary eye examination. The algorithms are also designed for diagnostic accuracy, which also have time saving abilities and increase the reach of care to a higher number of patients in the available time at the camp.

The advent of Artificial Intelligence has certainly proved to be extremely beneficial to detect diabetic retinopathy in patients at an early stage and also the advent of cloud to store Electronic Medical Records is another boon, says Dr Sheila. 

The software provided to the hospital is a retina risk assessment Software-as-a-Service platform available over the cloud. Patients’ images of the fundus of the eye are captured using smartphone cameras, handheld and table top fundus cameras. The images are then uploaded onto the Machine Learning-based platform, which can receive patient data through various sharing platforms of choice. 

The captured retinal images are analysed by the AI-based software that is programmed to detect diabetic retinopathy, age-related macular degeneration, glaucoma and various other retinal pathologies that require ophthalmologist’s intervention. These cutting-edge algorithms assess the eye condition on the severity and extent of damage to the eye. Based on this, optometrist at the camp site in consultation with the ophthalmologist at the base hospital, who has access to better ocular imaging equipment, decides if the patient has to necessarily travel to seek better healthcare services or is in emergency to save from loss of vision. Thus, AI, here, becomes mission critical in the time taken for decision-making as well as providing a cost-effective treatment to the patient. 

Scope for sophistication in AI application

Agreed that AI is a boon to aid faster diagnosis and early intervention, but Dr Sheila John believes that researchers working in this have just scratched the surface. There is a lot more to do in developing a 100% reliable software that can negate errors in diagnosis. Dr Sheila says she would any day rely on human decision-making skills rather than relying on a computer-based diagnostic decision. Also, she asks, “Who is going to take the liability for a wrong, unfortunate diagnosis? Should the ophthalmologist take the blame, or should the AI software be blamed? This blame game can cost the vision of a patient.”

In fact, Diabetic Retinopathy, itself, is a condition that has tremendous variations in the symptoms and presentations expressed by diabetics. Each patient will manifest different stages of the condition in many different ways at different times of testing. Thus, it is certain that the difficulties in diagnosis of DR is something AI-based software platforms should be designed to address. 

Dr Sheila, who bets on telemedicine, says as long as the 70% rural population are in need of quality healthcare, telemedicine will be critical in closing the gap by overcoming distance barriers through joint efforts of the government and private healthcare institutions. And in combination with digital technology tools, telemedicine can reshape the country’s healthcare industry’s strategies to better improve patient care.  

She says, India’s telemedicine now has a meaningful identity with the Prime Minister Narendra Modi-led union government releasing the Telemedicine Practice Guidelines on 25th March, 2020. She ascertains that telemedicine is an essential service, which enables timely access to appropriate interventions. In India, providing in-person healthcare is challenging, particularly given the large geographical distances and limited resources. Telemedicine’s major advantage is saving cost and time especially for rural patients. Similarly, Dr Sheila is looking forward to a policy on the ‘Ethical Use of AI in Healthcare’ that gives physicians more confidence to trust technology and also use digital tools in a more judicious manner. It is a must, by the government of India, to frame guidelines on the incorporation of AI in the Medical Equipment Regulatory Act, she says. 

So, the need for higher liability, accuracy, and ethical use of AI leaves tremendous room for research in this field. Cloud-based software with in-built AI and ML are the way forward for Diabetic Retinopathy diagnosis, providing scope for young researchers, both in healthcare and technology, to converge, discover, innovate, ideate, and fine-tune the software application to have better abilities in the diagnosis and treatment of Diabetic Retinopathy.  

Anusha Ashwin
Anusha Ashwin

When I started out as a trainee copy editor at a publishing house, least did I expect myself to be traveling along with the digital evolution of content. Ever since the digital model became a medium of providing content, my career kick-started. 
I consider my career to take a solid shape during the current phase, as I set shot to work in times of the digital era. Today, I am able to resonate with the demands of writing content tailored to meet the requirements of the new age consumer that is driven by SEO and SEM.
While with CyberMedia, I have handled content at premier magazines like BioSpectrum and Voice&Data – India’s foremost business magazines in the biotech & telecom verticals that have been instrumental in bridging the gaps in India’s life sciences and ICT domains respectively.
The blessing here is that, through the writings, I am on a constant self-discovery mode. I have found a passion, which is writing in focus on digital healthcare, communication-based tech startups, and health-tech entrepreneurs. Somewhere down me, there is this strong educational foundation in Microbiology and Biotechnology that had to play a part!
As India becomes more Atmanirbhar, I am destined to stay focussed on aligning my content contribution passion with Digital India plans, where I interact with numerous startup entrepreneurs and other organizations that are going to be part of the ambitious Make in India and Make for India programs.

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Scroll to Top
👋 Hello
Hello!! 👋 Manish here, Thanks for visiting The Healthcare IT Experts Blog !! How can i help you?