Abstract:
Diabetic retinopathy (DR) is the leading reasons for people losing their sight across the
world, particularly for those suffering from diabetes. Traditional diagnosis processes are
slow and rely heavily on human interaction which can lead to errors. Advances in
technology have made it possible to develop systems that automate processes, much like
the one this project aims to create. Through great research in neuroimaging data
now scan retinal images and
processing, CNNs (Convolutional Neural Networks)
categorize the extent of DR damage £4].
model as accuracy
can
This project will first integrate both AI image Processing and deep learning techniques
to achieve maximum accuracy when developing algorithms. After refining and testing
them extensively, Such evaluation with performance indicators will be used for the final
, sensitivity, specificity, and many more, by taking advantage of the
vast repositories available online. Some of the datasets provided online include the
EyePACS database on Kaggle and Messidor [2].
In the most effective and optimal use of this DR detection project, less populated regions
lacking ophthalmologists will have improved accuracy of diabetic patient treatment
prior to them reaching a stage needing eyesight surgery. With this non-invasive system,
patients will be able to receive timely treatment, greatly reducing die risk of blindness.
On a larger scope, abundant possibilities for deeper research will be provided
