Google’s AI program can detect diabetic eye disease
New Delhi: Google’s new research in machine learning using artificial intelligence can now detect diabetic retinopathy, a disease in which small blood vessels and neurons of the retina are damaged due to diabetes.
“Machine Learning can help doctors identify patients in need, particularly among under-served populations. One of the most common ways to detect diabetic eye disease is to have a specialist examine pictures of the back of the eye and rate them for disease presence and severity. Severity is determined by the type of lesions present (for example, microaneurysms, haemorrhages, hard exudates, etc.), which are indicative of bleeding and fluid leakage in the eye. Interpreting these photographs requires specialized training, and in many regions of the world, there aren’t enough qualified graders to screen everyone who is at risk,” said Varun Gulshan from Google.
Working closely with doctors both in India and the US, Google created a development data set of 128,000 images which were each evaluated by 3-7 ophthalmologists from a panel of 54 ophthalmologists. This dataset was used to train a deep neural network (a computer system modelled on the human brain and nervous system) to detect referable diabetic retinopathy.
Research results showed that Google’s algorithm’s performance was on par with that of ophthalmologists. The research has also been published in the latest Journal of the American Medical Association. Google has started using the technology in India at three eye hospitals—Aravind Eye Hospital in Madurai, Sankara Nethralaya in Chennai and Narayana Nethralaya in Bengaluru.
“Diabetic retinopathy is the fastest growing cause of blindness, if left untreated can lead to irreversible blindness. If we detect the disease on time, the treatment is easy. Moreover, not many people suffering from diabetes go for screening of diabetic retinopathy. The technology which is in its early stages is proving beneficial in screening these cases as there is a huge shortage of ophthalmologists capable of detecting the disease in India where diabetes is so prevalent,” said Dr R. Kim, chief medical officer at Aravind Eye Hospital.
“We are using it in our hospital but this is in initial stages. However, we may need further research to determine the feasibility of applying this algorithm in the clinical setting. It is also important to determine the use of this technology in outcomes compared with current ophthalmologic assessment,” Dr Kim said.
Diabetes specialists have welcomed the technology and are expecting better outcomes of the treatment. “This AI-aided eye pathology recognition would be of great use in India, where diabetic eye disease is largely neglected because of shortage of expert eye doctors and limited awareness of this morbid affliction in patients. This project could help diagnose retinopathy better thus permitting early treatment and avoidance of blindness,” said Dr Anoop Misra, chairman, Fortis-C-DOC Centre for Diabetes, Metabolic Diseases and Endocrinology.
According to the National Family Health Survey-4, conducted across 26 states and Union territories, the overall incidence of diabetes was found to be 20.3%.