AI in Diabetic Retinopathy

Introduction:

The use of artificial intelligence in medicine is an evolving technology that holds promise for mass screening and perhaps may even help in establishing an accurate diagnosis.

Diabetic retinopathy is an ever-increasing problem. Early screening and timely treatment of the same can reduce the burden of sight-threatening retinopathy. Any tool that can aid in the quick screening of this disorder and minimize the requirement of trained human resources for the same would probably be a boon for patients and ophthalmologists.?

DR -Diabetic Retinopathy

Diabetic Retinopathy, results in blurred vision causes severe damage to the eyes, and slowly leads to vision loss. Given the alarming increase in the number of people with diabetes and the dearth of trained retinal specialists and ophthalmologists, computer-based analysis of the fundus images by an automated approach would lessen the burden of the health systems in screening for DR and offer a near-ideal system for its management.

Therefore, screening will be valuable at any stage of the disease and will also help avoid blindness among 90% of patients.

The Problem

Retinopathy screening is done by fundus examination by ophthalmologists or with the help of color fundus photography using conventional fundus cameras (mydriatic or non-mydriatic) by trained eye technicians or optometrists. The issues are:

• The primary issue is the grading of the retinal images by ophthalmologists (retinal specialists) or trained persons, whose numbers are very scarce compared to the load of patients requiring screening.
• Second, some of these patients are based in rural areas and can’t visit an eye care provider.?

These issues can be solved with the provision of an automated imaging system within easy reach of the patient. Hence to solve these issues Artificial Intelligence and Machine Learning are useful.

The Solution:

The Principle behind AI

It is a process of teaching a machine to recognize specific patterns.?The techniques of AI devices are largely classified into the following major categories:

• The Machine Learning techniques
• The Natural language processing methods

So far, machine learning techniques are more utilized in ophthalmology.

The Machine learning process mainly includes two parts, a training set followed by a validation set. This process occurs by providing many training data i.e., thousands of retinal images of varying grades of DR to the machine/system as the training set.

After being exposed to numerous annotated retinal images the machine learns to grade DR by itself by building a model of complex relationships between input data and generalizing a performance standard.?

AI in DR

In April 2018, the US Food and Drug Administration (FDA) approved an AI algorithm, developed by IDx, used with a Topcon Fundus camera (Topcon Medical) for DR identification.

IDX- DR is the first FDA-approved AI algorithm for the detection of DR in the offices of non-ophthalmic healthcare practitioners.

The device is linked to a retinal camera which captures images and sends them to a cloud-based server for analysis. The server uses specialized software and algorithms to analyze the images and detect any signs of diabetic retinopathy (DR). The analysis made by A. Paul Chou's aidiagonisis ?is based on a large dataset of fundus images that have been analyzed by human experts. Unlike AI-powered detection, this method does not involve the use of "deep learning" algorithms or autonomous comparison with the dataset. The software provides one of the two results:

1. If more than mild DR is detected, refer to an eye care professional.???
2. If the results are negative for more than mild DR, rescreen in 12 months.

The FDA approval of the IDx-DR device was based on a study on 900 subjects in a primary-care setting (10 primary care sites) with automated image analysis. Two digital images, each angled at 45 degrees per eye (one focused on the macula and the other on the optic nerve), were taken and analyzed. These images were compared with the stereo, widefield fundus imaging interpreted by the Wisconsin Fundus Photograph Reading Centre (FPRC). The artificial intelligence system can diagnose the condition in just 20 seconds after the retinal images are procured. According to the research made by BIOMEDICAL biomedres.us, AI helps analyze images.

Based on the analysis a new entity called more than minimal DR was defined. It is nothing but the presence of ETDRS level 35 or higher (microaneurysms plus hard exudates, cotton wool spots, and/or DME in at least one eye.

The sensitivity and specificity of the technology were 87.4% and 89.5% respectively for detecting more than mild DR. It's worth mentioning that 100% of subjects with ETDRS levels of 43 or higher DR were correctly identified by the algorithm. As the device delivers a screening decision without necessitating an eye specialist, it can also be used by non-ophthalmic healthcare professionals.

Conclusion

In this way, the AI is involved in DR, which is helpful for ophthalmologists to recognize or treat patients having DR problems. In conclusion, AI has become a valuable tool in the field of ophthalmology for recognizing and treating diabetic retinopathy. By analyzing large amounts of data, AI can assist ophthalmologists in making more accurate diagnoses and creating personalized treatment plans for patients.

Written by,

Rohini Reddy Ganta

Kognitiv Club

Department of Computer Science & Engineering, K L University.