Stroke Prediction from Hypertensive Retinopathy

Hypertensive Retinopathy (HR) is known as the damage to the eye which occurs due
to high blood pressure. This HR may lead to permanent vision lost hence timely
diagnosis and treatment of this disease is very important. Fundus image analysis is
used to diagnose HR and stroke prediction. There are four steps in the proposed
systems which are the Image Enhancement, Fourier Fractal Dimension, Logistic
Regression Classifier and Stroke Prediction Model. The proposed system consists of
method used to analyse retina blood vessels using Fourier Fractal Dimension to
extract the complexity of the retina blood vessels enhanced in different scales.
Logistic regression was used to model the classifier for stroke prediction. The
probability from 0 to 0.5 was classified as control case and the probability from 0.5
to 1 was classified as stroke case. From 20 images used in this project only 14
images was classified as the stroke case. The estimated percentage increase in the
odds of incident of stroke is calculated and categorised according to the Hypertensive
Retinopathy stages.