m
Recent Posts
Connect with:
Thursday / May 19.
HomeminewsArtificial Intelligence Shows Potential to Fight Blindness

Artificial Intelligence Shows Potential to Fight Blindness

The use of artificial intelligence and deep learning technology to detect diabetic retinopathy is “new and exciting” according to Clinical Associate Professor Nitin Verma.

His comments are in response to a study published online in Ophthalmology, the journal of the American Academy of Ophthalmology. Researchers from the Byers Eye Institute at Stanford University described how they used deep-learning methods to create an automated algorithm to detect diabetic retinopathy.

“What we showed is that an artificial intelligence-based grading algorithm can be used to identify, with high reliability, which patients should be referred to an ophthalmologist for further evaluation and treatment,” said lead author, Dr. Theodore Leng. “If properly implemented on a worldwide basis, this algorithm has the potential to reduce the workload on doctors and increase the efficiency of limited healthcare resources. We hope that this technology will have the greatest impact in parts of the world where ophthalmologists
are in short supply.”

Another advantage is that the algorithm can be run on a common personal computer or smartphone with average processors.

Dr. Leng and his colleagues created an algorithm based on more than 75,000 images from a wide range of patients representing several ethnicities, and then used it to teach a computer to identify between healthy patients and those with any stage of disease, from mild to severe.

His algorithm could identify all disease stages, from mild to severe, with an accuracy rate of 94 per cent. It would be these patients that should see an ophthalmologist for further examination.

Potential for Universal Standardised Screening

Clin. Assoc. Prof. Verma said adapting artificial intelligence to achieve universal standardised screening of diabetic retinopathy would be key to early disease detection and management.

“Retinal photography is the backbone of any diabetic retinopathy screening program,” said Clin. Assoc. Prof. Verma. “Technology for retinal imaging has made great progress in terms of availability and affordability. Retinal cameras and optical coherence tomography (OCT) machines are installed in most optometric and ophthalmology practices in Australia. There is no standardisation of the grading of the characteristic changes seen on retinal photographs, which is a key step for implementing any of the referral guidelines in this context. Photos taken in different locations are read and graded by individuals with varying degrees of experience. A central reading centre, where retinal images are transmitted to for this purpose, is the logical answer to this problem. A few centers do exist in the country but are not being utilised for routine screening.

“To ensure standardisation, if everybody were to start sending their photos to a central/regional reading centre, it is likely that the system will be overwhelmed by the numbers of photos that need to be read. As screening rates increase, the number of photographs that need to be read will also increase substantially. For example, in the USA where there are about five million diabetics (over the age of 40), it is estimated that about 32 million photos would need to be evaluated (Wong and Bressler) if screening became universal.

“Deep learning technology and artificial intelligence is an ideal way of overcoming this problem and standardising the reading and grading of retinal images. A number of studies have demonstrated the high sensitivity, specificity and accuracy of deep learning technology in retinopathy screening. In addition, if the software was to be incorporated in the fundus camera itself (and ultimately in OCT machines which are being used more and more commonly in screening for diabetic macular edema), there would be no need for a centralised system and the recommendations would be part of a machine generated report.”

Clin. Assoc. Prof. Verma concluded, “The use of artificial intelligence and deep learning technology has been used for some time but adapting it to situations like diabetic screening is new and exciting and has the potential to make the goal of universal standardised screening for diabetic retinopathy a reality in the not too distant future.”

Approval from the US Food and Drug Administration is required before the algorithm can be used in patients on a broad basis.

mivision’s education article on page 59, written by Dr. Swetha Jeganathan and Clinical Associate Professor Nitin Verma puts the spotlight on diabetes mellitus, examining the disease epidemiology and ocular manifestations.

DECLARATION

DISCLAIMER : THIS WEBSITE IS INTENDED FOR USE BY HEALTHCARE PROFESSIONALS ONLY.
By agreeing & continuing, you are declaring that you are a registered Healthcare professional with an appropriate registration. In order to view some areas of this website you will need to register and login.
If you are not a Healthcare professional do not continue.